blending efficiency of asphalt mixtures containing
TRANSCRIPT
Blending Efficiency of Asphalt Mixtures Containing
Recycled Asphalt Pavement and Shingle
Submitted to the Tennessee Department of Transportation
Research Development and Technology Program
Baoshan Huang, Ph.D., P.E.
Xiang Shu, Ph.D.
Sheng Zhao, Ph.D.
Department of Civil and Environmental Engineering
The University of Tennessee, Knoxville
March 2017
Technical Report Documentation Page
1. RES2013-32 2. Government Accession # 3. Recipient’s Catalog #
4. Title & Subtitle
Blending Efficiency of Asphalt Mixtures Containing Recycled Asphalt Pavement
and Shingle
5. Report Date
March 2017
7. Author(s)
Baoshan Huang, Xiang Shu and Sheng Zhao
6. Performing Organization Code
9. Performing Organization Name & Address
The University of Tennessee
Depart of Civil and Environmental Engineering
325 John D. Tickle Building
Knoxville, TN 37996
8. Performing Organization Report #
10. Work Unit # (TRAIS)
12. Sponsoring Agency Name & Address
Tennessee Department of Transportation
Materials and Test Division
11. Grant #: 36B268 (FHWA)13. Type of Report & Period Covered
14. Sponsoring Agency Code
15. Supplementary Notes
16. Abstract
This study aims to address blending issues in warm mix asphalt (WMA) and hot mix asphalt (HMA) containing recycled asphalt
pavement (RAP) and recycled asphalt shingle (RAS). Two fingerprinting methods of asphalt covered in this study are gel permeation
chromatography (GPC) and atomic force microscopy (AFM). Laboratory testing methods for determining blending efficiency employed
in this study include determining aged binder mobilization rate and staged extraction method. The factors affecting the blending
efficiency include mixing time, mixing temperature, aged binder content and effect of WMA technology. Major conclusions from the
study are summarized as follows:
1) There existed a strong correlation between the percentage of large molecular size (LMS) and the complex modulus (G*) of asphalt
binder based on the comparison of GPC and dynamic shear rheometer (DSR) test results. As mixing time and temperature increased,
more blending occurred in the RAP/RAS mixture. The size of virgin aggregate did not affect the blending efficiency of RAS in
pavement mixtures.
2) “Blending Charts” could be generated between the RAP/RAS binder content in the blend with the large molecular size (LMS%)
defined by molecular weight and the relations were found to be linear. The mobilization rate of RAP binder decreased with the increase
of the RAP percentage in the mixture. RAS binder mobilization rate increased with RAS percentage growing from 2.5% to 5%, but
decreased with RAS percentage passing 5%. The highest mobilization rate was around 61% and found on 5% RAS mixture while the
mobilization rate of mixture containing 10% RAS could be as low as 36%.
3) Trichloroethylene (TCE) was found to be the most effective solvent used in the study for staged extraction. The binder coating on the
raw RAP and RAS aggregates was proved to be homogeneous and the layer stripping did occur in a well-controlled composite binder
system. Based on well-designed staged extraction and GPC analysis, binder film coating virgin aggregates was approximately
homogeneous in RAP mix, while a non-homogeneous binder was generated on RAP aggregates. A potential composite binder system
was found coating the virgin aggregates in RAS mix. The diffusion study shows that within the mixture storage time, binder diffusion
can be accomplished in both warm and hot mixes containing RAP, indicating old binder mobilization, rather than binder homogeneity,
could be more critical in RAP mix. The binder diffusion in RAS mix was captured in a very slow rate.
4) WMA additives yielded higher blending ratio than control mix produced at 135C, but the temperature of 165C still produced the
mix with the highest blending ratio value. This indicates that a concern still exists over asphalt blending even if WMA additives are
used. Foaming technology yielded a higher blending ratio, indicating foamed WMA may yield a higher blending than regular HMA. It
was also found that temperature rather than coating is more critical in RAS blending.
5) According to the observations, AFM proved to be capable of differentiating virgin binder from RAS binder in terms of
microstructures. The microstructures of tear-off RAS binder was found to be temperature-dependent, but changed very little within the
range from 60C to 180C. Virgin binders selected in this study could not blend through a RAS binder layer of 300 μm within 30
minutes at 180C. On the basis of observations on the interfacial zone, RAS binder was found to be “mixing” but not “blending” in a
mixing zone of 25 to 30 μm.
17. Key Words
asphalt mixture, blending efficiency, recycled asphalt pavement,
recycled asphalt shingle, GPC, AFM
18. Distribution Statement
No restrictions. This document is available to the public
through the National Technical Information Service,
Springfield, VA 22161
Security Classification (of this report)
Unclassified
Security Classification (of this
page)
Unclassified
21. # Of Pages 22. Price
Form DOT F 1700.7 (5-02) Reproduction of completed page authorized
186 $215,282.05
DISCLAIMER
This research was funded through the State Planning and Research (SPR) Program by the Tennessee
Department of Transportation and the Federal Highway Administration under RES2013-32: Blending
Efficiency of Asphalt Mixtures Containing Recycled Asphalt Pavement and Shingle.
This document is disseminated under the sponsorship of the Tennessee Department of Transportation and
the United States Department of Transportation in the interest of information exchange. The State of
Tennessee and the United States Government assume no liability of its contents or use thereof.
The contents of this report reflect the views of the author(s) who is(are) solely responsible for the facts
and accuracy of the material presented. The contents do not necessarily reflect the official views of the
Tennessee Department of Transportation or the United States Department of Transportation.
iii
Acknowledgements
We would like to begin by thanking the Tennessee Department of Transportation
(TDOT) for funding this research project. In this research, we have continued to
collaborate closely with regional engineers and local technicians at the TDOT Materials
and Test Division and local asphalt plants. They have provided valuable support towards
the fulfillment of the research objectives. Without their support, it would be impossible
for us to finish this research project. We would also like to thank the administrative staff
from the TDOT research office who have worked very closely with our research team and
kept the whole project on the proposed schedule.
iv
Executive Summary
The current tendency in paving industry is to increase the use of recycled asphalt
pavement (RAP) and recycled asphalt shingle (RAS). TDOT Materials and Test Division
has been collaborating with the University of Tennessee conducting studies of recycling
RAP/RAS into hot-mix asphalt (HMA) or warm mix asphalt (WMA). The studies have
shown that one of the big advantages of WMA is its capability of incorporation of higher
percentages of RAP/RAS. However, one of the reasons that limit the high recycled
amount is the unknown blending between virgin and RAP/RAS binders. This study aims
to address blending issues in WMA and HMA containing RAP and RAS. Major research
activities are summarized as follows:
1) Two fingerprinting methods of asphalt covered in this study include gel
permeation chromatography (GPC) and Atomic Force Microscopy (AFM). GPC
was used to determine aged binder concentration in and blend. AFM was used to
directly detect the blending process between aged binder and virgin binder.
2) Laboratory testing methods for determining blending efficiency employed in this
study include determining aged binder mobilization rate and staged extraction
method. Using these testing methods, the factors affecting the blending efficiency
include mixing time, mixing temperature, aged binder content and WMA
technology. The effects of these factors were investigated in this study.
Based on the results from the study, the following conclusions can be drawn:
1) As mixing time and temperature increased, more blending occurred in the
RAP/RAS mixture. The size of virgin aggregate did not affect the blending
efficiency of RAS in pavement mixtures. The most efficient blending of RAS may
occur at approximately 5% RAS content. There existed a strong correlation
between the percentage of large molecular size (LMS) and the complex modulus
v
(G*) of asphalt binder based on the comparison of GPC and the dynamic shear
rheometer (DSR) test results.
2) A new LMS% related to molecular weight distribution derived from GPC analysis,
was developed to differentiate the RAP/RAS and virgin binders as well as their
blends. “Blending Charts” could be generated between the RAP/RAS binder
content in the blend with the newly defined large molecular size (LMS%) and the
relations were found to be linear. The results generated from “ blending charts”
shows that RAP binder mobilization rate decreased with the increase of the RAP
percentage in the mixture with mobilization rates close to 100% at low RAP
mixtures (10% and 20%), but dropping from 73% to 24% with RAP percentage
varying from 30% up to 80%. RAS binder mobilization rate increased with RAS
percentage growing from 2.5% to 5%, but decreased with RAS percentage
passing 5%. The highest mobilization rate was around 61% and found on 5%
RAS mixture while the mobilization rate of mixture containing 10% RAS could
be as low as 36%.
3) It was found that trichloroethylene (TCE) was the most effective solvent used in
the study for staged extraction. The binder coating on the raw RAP and RAS
aggregates was proved to be homogeneous and the layer stripping did occur in a
well-controlled composite binder system. A well designed step-extraction method
with progressive wash times could replace equal-time extraction method, and
yielded better analysis. Based on well-designed staged extraction and GPC
analysis, it was found that, in RAP mix, binder film coating virgin aggregates was
approximately homogeneous, while a non-homogeneous binder was generated on
RAP aggregates. A potential composite binder system was found coating the
virgin aggregates in RAS mix. The diffusion study shows that within the mixture
storage time, binder diffusion can be accomplished in both warm and hot mixes
containing RAP, indicating old binder mobilization, rather than binder
vi
homogeneity, could be more critical in RAP mix. The binder diffusion in RAS
mix was captured in a very slow rate. It was suggested that old binder activation
and binder homogeneity can both be critical for RAS mix.
4) WMA additives yielded higher blending ratio than control mix produced at 135C,
but the temperature of 165C still produced the mix with the highest blending
ratio value. Foaming technology yielded a higher blending ratio, indicating
foamed WMA may yield a higher blending than regular HMA. It was also found
that temperature rather than coating is more critical in RAS blending. Finally, the
mix produced with coarse virgin aggregates and medium RAP may not be
sensitive enough to test the effect of WMA additives on blending, while the mix
with medium virgin aggregates and fine RAP was more effective.
5) The blending of virgin-RAS binder was first observed in this study by AFM.
According to the observations, AFM proved to be capable of differentiating virgin
binder from RAS binder in terms of microstructures. The microstructures of
tear-off RAS binder was found to be temperature-dependent, but changed very
little within the range from 60C to 180C. Virgin binders selected in this study
could not blend through a RAS binder layer of 300 μm within 30 minutes at
180C. On the basis of observations on the interfacial zone, RAS binder was
found to be “mixing” but not “blending” in a mixing zone of 25 to 30 μm.
Recommendations
On the basis of the conclusions obtained in this study, the following
recommendations can be made:
1) Better tracking materials, other than the round aggregate, could be used for the
quantitative evaluation of the old binder mobilization rate. The texture, size
and other surface properties of the tracking materials could be considered as
influencing factors.
vii
2) It is recommended that new methods be developed to quantify the binder
homogeneity through AFM. Statistical methods could be used to track the
numerical change of the domains. Layered system with well controlled binder
film thickness is recommended to characterize the diffusion coefficient of the
virgin binder through the old binder.
3) Neutron scattering are also recommended for use in blending research. Samples
with different blending degrees may express different inner structural properties
and might be revealed by neutron scattering detection. Additionally, neutron
scattering samples will not go through any destructive damage during preparation
and testing processes.
4) Test methods designed in this study have been proved to be useful in laboratory
mixing. However, their effectiveness in plant mixtures still need to be further
verified. In the follow-up study, the test methods for blending efficiency may be
modified according to needs in future plant mixture study.
viii
Table of Contents
CHAPTER 1 INTRODUCTION ............................................................................ 1
1.1 Problem Statement .................................................................................... 1
1.2 Objectives ................................................................................................. 2
1.3 Scope of Study .......................................................................................... 2
CHAPTER 2 LITERATURE REVIEW .................................................................. 4
2.1 Blending Efficiency Studies...................................................................... 4
2.2 Gel Permeation Chromatography (GPC) .................................................. 6
2.3 Atomic Force Microscopy (AFM) .......................................................... 10
CHAPTER 3 RESEARCH METHODOLOGY ................................................... 13
3.1 Fingerprinting Methods .......................................................................... 13
3.1.1 Gel Permeation Chromatography (GPC) ..................................... 13
3.1.1.1 5/13 Method ...................................................................... 14
3.1.1.2 Molecular Weight Method ................................................ 15
3.1.2 Atomic Force Microscopy (AFM) ............................................... 19
3.2 Laboratory Methods to Determine Bleeding Efficiency ......................... 22
3.2.1 Building “Blending Chart” .......................................................... 22
3.2.1 Aged Binder Mobilization Rate ................................................... 23
3.2.3 Staged Extraction Method ............................................................ 26
3.2.3.1 Solvent Selection .............................................................. 27
3.2.3.2 Selective Dissolution ........................................................ 29
3.2.3.3 Binder Homogeneity of RAP and RAS Particles .............. 32
3.3 Factors Affecting the Blending Efficiency .............................................. 33
3.3.1 Mixing Factors ............................................................................. 33
3.3.1.1 RAP ................................................................................... 33
ix
3.3.1.2 RAS ................................................................................... 35
3.3.2 Effect of Aged Binder Content ..................................................... 39
3.3.3 Effect of WMA technologies ....................................................... 41
3.3.3 Stage Extraction Method .............................................................. 45
3.3.3.1 Effect of Extraction Time .................................................. 45
3.3.3.2 Diffusion Studies............................................................... 48
CHAPTER 4 LABORATIORY TEST RESULTS AND DISCUSSION .............. 53
4.1 Blending Chart ........................................................................................ 53
4.2 Factors Affecting the Blending Efficiency .............................................. 54
4.2.1 Effect of Mixing Factors .............................................................. 54
4.2.1.1 RAP ................................................................................... 54
4.2.1.2 RAS ................................................................................... 62
4.2.2 Effect of Aged Binder Content ..................................................... 66
4.2.3 Effect of WMA technology .......................................................... 68
4.2.3.1 RAP ................................................................................... 71
4.2.3.2 RAS ................................................................................... 76
4.3 Results from Extraction Method .......................................................... 78
4.3.1 Validation Results of Staged Extraction....................................... 78
4.3.2 Effect of Extraction Time ............................................................. 79
4.3.3 Diffusion Studies.......................................................................... 80
4.4 AFM ........................................................................................................ 86
4.4.1 Temperature Dependence of Microstructures in RAS Binder ..... 91
4.4.2 Blending Degree between RAS and Virgin Binder...................... 92
4.4.3 Scanning on the Interfacial Zone ................................................. 96
CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS ........................ 100
Recommendations ....................................................................................... 103
REFERENCES ................................................................................................... 172
x
APPENDICESAPPENDIX A: GPC Chromatograms ........................................ 105
1
CHAPTER 1 INTRODUCTION
1.1 Problem Statement
This proposed research project is the continuation of the TDOT-sponsored study
entitled “Laboratory and Field Evaluation of Warm Mix Asphalt Pavement in Tennessee”.
TDOT Materials and Test Division has been collaborating with the University of
Tennessee conducting studies of recycling recycles asphalt pavement (RAP) and/or
recycled asphalt shingle (RAS) into hot-mix asphalt (HMA) or warm mix asphalt (WMA).
The studies have shown that one of the big advantages of WMA is its capability of
incorporation of higher percentages of RAP/RAS. However, when RAP/RAS is recycled
into WMA or HMA mixtures, one key question needs to be answered: how much of the
aged asphalt binder in RAP and/or RAS can be blended into virgin asphalt? What is the
role of RAP/RAS in asphalt mixtures? Do they act as black rock or the aged binder can
be fully blended into virgin binder? To fully understand the role of RAP/RAS in asphalt
mixtures and to maximize the use of RAP/RAS in WMA, these questions have to be
answered.
In addition to blending efficiency of RAP and RAS, one particular concern
regarding RAP limits in current TDOT specification will be identified and explored.
Specifically, TDOT specification allows higher percentages of RAP in mixtures made
with non-modified asphalt binders (PG 64-22 and PG 67-22), which are mainly used in
relatively thin pavement structures in state routes, whereas higher restrictions are applied
to the mixtures made with modified-asphalt binder (such as PG 76-22), which are
typically used for thick pavement structures such as interstates. The performance of these
two types of pavements with respect to RAP content is of great concern and interest to
TDOT and will provide a sound basis for more reasonable limitations and restrictions of
2
RAP usage. Currently, recycle of RAS into asphalt mixtures is gaining more and more
attention in asphalt paving industry. The blending efficiency issue is of paramount
importance to highway agencies.
This research will significantly benefit the economy of the State of Tennessee
through use of both WMA and RAP technologies in flexible pavements:
(1) Less fuel consumption to heat aggregates due to lowered mixing and
placement temperatures;
(2) Reduced emission and improved working environments;
(3) More RAP/RAS may be recycled into asphalt mixtures;
(4) Significant cost saving for asphalt mixtures;
(5) More asphalt pavements can be rehabilitated or built with the same budget;
(6) Beneficial to the environments in Tennessee due to the use of WMA and
RAP/RAS technologies.
1.2 Objectives
The objectives of the proposed research are to:
(1) Fingerprint the liquid virgin asphalt, RAP, RAS, and recycled asphalt
mixtures containing RAP/RAS using available chemical testing techniques.
(2) Develop a laboratory testing method to determine the blending efficiency of
recycled asphalt mixtures containing RAP/RAS.
(3) Investigate the effects of different blending efficiencies on the laboratory
performance of recycled asphalt mixtures containing RAP/RAS.
1.3 Scope of Study
The scope of the research work includes:
3
To complete a synthesis of literature review and state DOT survey on the use
of RAP and/or RAS in asphalt mixtures;
To fingerprint the liquid virgin asphalt, RAP, RAS, and recycled asphalt
mixtures used by TDOT with available chemical testing techniques;
To develop a practical procedure to determine the blending efficiency of
RAP/RAS in recycled asphalt mixtures;
To investigate into different factors affecting blending efficiency and
subsequently their effects on laboratory performance of recycled asphalt
mixtures.
4
CHAPTER 2 LITERATURE REVIEW
2.1 Blending Efficiency Studies
A major cause of concern in the reuse of asphalt binder from a reclaimed asphalt
source such as recycles asphalt pavement (RAP) and/or recycled asphalt shingle (RAS) is
whether or not the reclaimed binder truly blends with the virgin binder. “Black rock”
theory is based on the premise that the RAP may actually perform as nothing more than
an aggregate. While this may provide some use of the RAP as a recycled aggregate, the
major economic savings come from the reuse of the RAP binder. However, if the stiff,
highly oxidized binder does not truly blend with the virgin binder in the mixture, the use
of blending charts commonly used in pavements where RAP is included is unnecessary
[1, 2].
Blending charts are commonly used in practice as a guide to safe addition of RAP.
The premise of these charts is that up to a certain percentage of RAP can be added
without needing to decrease or “bump” the binder performance grade in order to
counteract the stiff, oxidized RAP binder [3]. Common practice is to add up to 15% RAP
without changing binder grade and between 15-25% RAP by only reducing one binder
grade with respect to both high and low temperatures. Beyond 25% RAP a blending chart
is required to estimate the binder grade necessary to create a mixture that will perform to
SuperPave standards. The use of blending charts in current practice was further supported
by NCHRP 9-12 conducted by McDaniel et al. [2] which concluded that partial blending
does occur to a significant extent. In order to further refute the “black rock” theory,
Soleymani et al [4] simulated a “black rock” condition. The researchers created three
mixtures, denoted Case A, Case B, and Case C. The black rock scenario, Case A, was
created by extracting the binder from a RAP and adding only the recovered aggregate to a
5
new mixture of virgin aggregate and virgin binder. Case B is considered a “true” mixture,
where the RAP is added directly to the mixture with the virgin aggregate and virgin
binder. The “total blend”, Case C, was created by mechanically blending the virgin binder
and recovered RAP binder and adding it to the virgin and RAP aggregate to make the
mixture. All three cases were evaluated using the SuperPave Shear Tester (SST) with
frequency sweep, simple shear, and repeated sheer at a constant height testing scenarios.
The researchers evaluated compared the results of the three cases statistically and found
that at a 10% RAP content none of the cases were differentiable. However, at 40% RAP
content, 45% of the “true” RAP mixtures (Case B) performed similarly to the 100% blend
(Case C), while only 5% performed similarly to the “black rock” scenario (Case A). The
remaining 50% of the mixtures didn’t not perform similarly to either Case A or Case C.
This study concluded that RAP does not behave as a black rock considering at a high
RAP content (40%) the mixture performed similarly to that of a 100% blend. The
researchers also state conclude that at least partial blending occurs in almost all cases.
Additionally, an important conclusion is that the suggestion that up to 15% RAP can be
added without a need to resort to blending charts is reasonable on the basis of this study.
Other attempts to address the blending efficiency question have been made.
McDaniel et al. [5] address this issue by using the “Bonaquist Approach” [6] which
considers pavement volumetrics and their relationship to the dynamic modulus (E*) of
the mixture. The Hirsch Model is used to create an estimated E* master curve.
Twenty-four mixtures were considered utilizing multiple contractors. Mixtures were
created mixes using a virgin Performance Grade (PG) 64-22 asphalt binder with 0%,
15%, 25%, and 40% RAP contents. Additionally, a softer PG58-28 virgin binder was used
at 25% and 40% RAP content. Each of the contractors made efforts to keep the mixture
gradation as close as possible to their counterparts. The binder was recovered using
n-propyl bromide and the complex modulus (G*) was investigated using the Dynamic
Shear Rheometer (DSR) and low temperature performance was tested using the Bending
6
Beam Rheometer (BBR). These values were used in the Hirsch Model to estimate the E*
master curve on the assumption that the recovered asphalt could be considered a 100%
blend between the virgin binder and RAP binder. The E* master curve was then generated
in the lab through performance testing for each of the mixtures. If the estimated E*
matched the mixtures true E*, the mixtures were considered to be 100% blended. Out of
21 samples containing RAP only three were considered to have poor blending and one
had partial blending.
A similar study was conducted by Mogawer et al. [7] looking specifically at 9.5
mm and 12.5 mm SuperPave mixtures. These mixtures comprised of PG58-22 binder
containing 30% and 40% RAP, PG64-22 binder containing 0%, 20%, 30%, and 40%
RAP, and PG52-34 and PG64-28 binder mixtures containing 0%, 20%, 30%, and 40%
RAP. Samples were collected from New Hampshire, Vermont, and New York. The
Christenson-Anderson model was used to develop a master curve for the extracted and
recovered binders from each mixture. These estimated master curve was then compared
to the E* master curve of the mixture to determine the degree of blending. A conclusion
of this study focused on the handling, mixing, and storage of the material in terms of its
impact on the mixture stiffness and cracking properties. The authors state that blending
efficiency seemed to be impacted by the discharge temperature of the mixture from the
silo.
2.2 Gel Permeation Chromatography (GPC)
Many works have been conducted that show the relationship between the
chemistry chemical composition and the mechanical behavior of the asphalt binder. GPC
is an analytical chemistry technique that yields the molecular weight distribution of a
given medium in solution. Figure 2-1 shows a simplified example of asphalt binder in
solution passing through a single multi-pore GPC column.
7
Figure 2-1 Example of Gel Permeation Chromatography
In Figure 2-1 the same column is represented four times at different elution
intervals (t = 0, 5, 8.5, and 11.25 minutes) as shown on the x-axis. Time can be directly
correlated to specific molecular weights on the basis of polymer standards such as
polystyrene. The y axis is representative of the refractive index (RI) output in mV. An
increase in refractive index indicates and an increase molecules of a given size, with the
maximum occurring where the most molecules of a given size are present. The example
is based on the assumption that asphalt binder can be simply separated into two fractions;
asphaltenes which are high in molecular weight and maltenes which are low in molecular
weight. At t=0 the asphaltenes and maltenes are injected in one solution. However, as
they pass through the column as represented by t=5 they begin to separate. This is due to
the absorption of smaller molecules by small pores in the columns packing. Larger
molecules are able to pass these small pores, thus causing separation between the
different molecular sizes. At t=8.5 the high molecular weight asphaltenes begin to exit the
column and a slight increase in the RI occurs. At t=11.25 the asphaltenes have completely
exited the column and the maltenes begin to exit. It is important to note that the peak of
the chromatogram is at t=11.25, indicating that there are more molecules present of the
8
correlating molecular size to t=11.25, which also corresponds to the region encapsulated
by the maltenes fraction.
Much work has been conducted studying the application of GPC to asphalt binder
chemistry. The following Table 2-1 outlines many uses and findings:
Table 2-1 Overview of literature involving the application of GPC to asphalt binder
Citation GPC Application
McCann et al. (2008)[8] Detected polymers in asphalt binders using
GPC
Lu and Isacsson (2002) [9]
Siddiqui and Ali (1999) [10]
Kim and Burati (1993) [11]
Churchill et al. (1995) [12]
Lee et al. (2011)[13]
Studied effects of aging on asphalt binder using
GPC
Snyder (1969)[14]
Determined asphalt molecular weight
distributions; based on polystyrene standard
estimated between 700-2400 Daltons
Gilmore (1983)[15] Identified the presence of antistripping agents
within asphalt cement
Kim et al. (2006)[16] Estimated RAP binder viscosity with GPC
Shen et al. (2007)[17] Studied effect of crumb rubber modification of
binders
Kim et al. (2013)[18]
Examined oxidative aging on
polymer-modified asphalt mixtures made with
Warm Mix Asphalt (WMA) technologies
Jennings et al. (1980, 1985)[19, 20]
Yapp, Durrani, and Finn (1991)[21]
Studied pavements ranging from good to poor
condition and compared the chromatogram;
established a tipping point within the
chromatogram at which pavement failure
occurs
Zhao et al. (2013)[22] Correlated an increase in large molecular sizes
to the complex modulus (G*) of binder
Many of the summarized works in Table 2-1 employed a measure of the Large
Molecular Size (LMS%) to compare the binders. The percentage of LMS% grows as the
9
binder is oxidized. Kim et al [16] correlated the percentage of LMS% directly to the
binder viscosity, as stated above. This important finding relates LMS% to the binder
viscosity, a performance property of asphalt paving. The LMS% is defined by the first 5
slices of the GPC chromatogram if the total chromatogram was divided into 13 equal
slices on the basis of elution time as shown by Figure 1.56. The LMS% is defined with
the following equation:
(2-1)
Zhao et al. [22] used LMS to estimate the complex modulus (G*) of asphalt
binder that contained RAS successfully. The researchers found that the complex modulus
increased as the LMS% increased. Correlating these properties is important in further
correlating the rheological properties of the asphalt binder to changes in the molecular
weight distribution.
In 1980 Jennings et al. [19] conducted research identifying a number of roadways
in Montana and ranking them on the basis of the amount of cracking the pavement had in
certain age ranges. If a pavement was less than 10 years of age and exhibited extensive
cracking it was ranked as “bad”, newer than 10 years old and only had some cracks it was
listed as “poor”, newer than 14 years with few cracks was considered “good”. Pavements
that were considered “excellent” were those that were 14 years or older and had little
cracking. Jennings noticed that as the ranking of the pavement declined (i.e. went from
“excellent” to “bad”) the LMS% of the chromatogram continued to increase. However,
Jennings also realized that the performance of the pavement was not based on asphaltene
content, which generally takes up most of the LMS% region.
Daly [23] fractionated asphalt binder with heptane and then tested each fraction in
the GPC compared to the total chromatogram. The asphaltene fraction is heptane
insoluble, while the maltene fraction is soluble. Daly found that the asphaltene fraction
was responsible for the LMS% region, however it also tailed well into the maltene
10
fraction. The maltene fraction was responsible for the remaining portion (e.g. the
distribution after the LMS) of the chromatogram. In the case that the asphaltene fraction
was solely responsible for the formation of cracks (i.e. stiffening of the binder) the
asphaltene fraction would not drag on into the medium and small molecular size regions.
There are some perceived limitations of this methodology. Namely, when GPC is
being used to identify molecular weight distributions the molecular weights are
qualitative, not quantitative. Thus, as previously shown, to quantitatively analyze data the
data needs to be normalized by dividing the area of the LMS% region by the area of the
total chromatogram. Furthermore, in effort to compare LMS % between samples the
researcher must compare all samples with the same base limits. RAP for example
typically will have more large molecules and those molecules will likely begin to elute
from the column before the molecules of the virgin binder. Thus the minimum limits in
which the area of the chromatogram is being analyzed need to be adjusted to meet those
of the RAP (e.g. shifted from a minimum of 8.7 minutes for virgin binder to 8.5 minutes,
which may be where the RAP begins to elute).
2.3 Atomic Force Microscopy (AFM)
Loeber’s group [24, 25] first investigated the bitumen heat-cast film using AFM,
and observed the well-known “bee” shaped microstructures with several micrometers in
diameter and tens of nanometers. Several years later, Pauli et al. [26] acquired the same
bee-shaped microstructures and advanced to correlate the “bees” with the amount of
asphaltenes in the binder by imaging solvent-cast film. Having found the same randomly
distributed bee-shaped structures, Jäger et al. [27] furthered the research and identified
four phases in topographic images of the bitumen (hard-bee, soft bee, hard matrix and
soft matrix), separating the higher and lower parts of the “bees” and the surrounding
phase. Later on, a more extensive research including 13 bitumen was conducted by
11
Masson et al. [28] with phase-detection mode in AFM. The similar phases as identified in
earlier research [27] were observed but named differently, with one new salt-like phase
found and termed sal phase. The authors also classified bitumen in three groups: one that
showed a fine dispersion (0.1 to 0.7 µm) in a homogenous matrix, one that showed
domains of about 1 µm and one that showed up to the aforementioned four phases. An
interesting finding in this research was that poor correlation was reported between the
asphaltene content and the bee-shaped structures. This finding was subsequently verified
by Pauli et al. [29], who first claimed the correlation between bee-shaped microstructures
and asphaltene content. Pauli’s group found ‘bees’ in maltenes without any asphaltenes
but no similar microstructure existing in de-waxed bitumen. This observation led to a
corrected statement that the microstructuring in bitumen, including well-known
bee-shaped microstructures, resulted from the interaction between the crystallizing
paraffin waxes.
Besides chemical composition of bitumen, temperature was another factor found
responsible for its microstructural change. Das et al. [30] studied the influence of
temperature on microstructures in bitumen by combining differential scanning
calorimetry (DSC) and AFM. It was found that the appearance of microstructures is
always in the crystallization temperature range of the same bitumen, while the dissolving
of these microstructures is related with the melting temperature range. Another study
addressing the similar issue was conducted by characterizing the microstructure for
various thermal scenarios like cooling or heating in a fast or gradual manner [31]. The
major findings of this research can be summarized as: microstructure possessed memory
of its previous thermodynamic state; elliptical domains (“bees”) showed the tendency to
orient relative to each other with the change of temperature; the microstructural
properties were found to depend on the maximum hold temperature of the bitumen.
Generally, these studies showed that AFM proved to be capable of fingerprinting
bitumen from different sources under certain thermodynamic condition, although the
12
mechanism of the development of microstructures still needs to be answered. This
motivated applying AFM to other areas of asphalt research, such as aging [32] and
moisture damage [33]. These published AFM applications, together with current need of
direct detection on the binder, which was previously discussed, incented using this
powerful microscopic tool to research on binder blending. The authors of this paper
recently evaluated the interaction and extent of blending between RAP-binder and virgin
binder by studying the microstructures of the ‘blending zone’ by using AFM [34]. The
blending zone was estimated to be about 50 µm. The blending zone can be considered to
be a completely blended ‘new material’, having an intermediate microstructural property
of the two materials. A design formula was proposed, which correlated the blending zone
dimension to temperature and mixing time. Since the concern over binder homogeneity is
more significant in RAS mixtures, blending between RAS binder and virgin binder was
evaluated in this study.
13
CHAPTER 3 RESEARCH METHODOLOGY
3.1 Fingerprinting Methods
3.1.1 Gel Permeation Chromatography (GPC)
When GPC works, analytes are dissolved in the solvent, commonly
tetrahydrofuran (THF), then stirred and injected into the columns that are packed with
different pore sizes. It can be seen in Figure 2-1, larger molecules escape from the pores
while smaller ones enter the pores easily with increased retention time. A differential
refractive index (RI) is used as the detector, since it is most commonly used for asphalt
chromatography.
GPC has been introduced into the blending efficiency study because it is capable
of differentiating aged binder from virgin binder due to the fact that aged binder has a
higher portion of large molecule than virgin binder. Figure 3-1 shows the results
comparing the GPC chromatograms of one virgin binder (PG 64-22), binder extracted
and recovered from one RAP and one RAS provided by TDOT, respectively. Each binder
gives a different LMS% defined as the ratio of area of the large molecules to the total area.
It can even be visually detected that the LMS% vary significantly from virgin binder to
recycled binder. The significant change in molecular distribution is the foundation of
using GPC in blending efficiency research.
14
Figure 3-1. Comparison of GPC chromatograms of virgin, RAP and RAS binders
3.1.1.1 5/13 Method
The concept of LMS increase with respect to the stiffening of asphalt binder due
to oxidation and aging is a popular one. The following table (Table 3-1) shows a number
of the different approaches to LMS%. Each argues that the chromatogram needs to be
divided into slices and then integrated, but there is variation on the number of slices that
need to be used. One could theoretically use a range of molecular weights to divide the
chromatogram. Evaluating the chromatogram on the basis of integration is straight
forward, and a vast majority of the literature that studies similar GPC applications to the
study of asphalt binder oxidation have been proven statistically significant with the use of
an integrated LMS% method. Our research group correlated the complex modulus, G*, of
virgin asphalt binder blended with RAS binder to the LMS % using the “5/13 method”.
“5/13 method” means the LMS% is defined by the first 5 slices of the GPC
chromatogram if the total chromatogram was divided into 13 equal slices on the basis of
15
elution time. It was proved that “5/13 method” correlated very well with the G* of
random combination of virgin and RAS binders (Figure 3-2), which means this method
can be used in further blending efficiency research.
Table 3-1. Differing opinions on LMS percentage
Researchers # Slices How many LMS
Asi, Al-Dubabi (1997) 12 n/a
Kim et al. (2006) 13 1-5
Churchill, Amirkhanian, Burati (1995) 10 n/a
Lee, Amirkhanian, Shatanawi (2006) 13 1-5
Doh, Amirkhanian, Kim (2008) 13 1-5
a) Tested at 25C b) Tested at 64C
Figure 3-2. The correlation between G* and LMS%
3.1.1.2 Molecular Weight Method
Since GPC columns need to be calibrated by standard solutions, mostly
polystyrene for asphalt research, the retention times in the chromatogram can be
converted to molecular weights using the calibration curve obtained using polystyrene
standards. Figure 3-3 shows the same chromatogram presented in Figure 3-1 but
re-arranged based on polystyrene molecular weights in Daltons. The fractions generated
by components with molecular weights less than 200 Daltons have been removed due to
y = 2E+08x + 262169 R² = 0.8681
0.0E+0
5.0E+6
1.0E+7
1.5E+7
2.0E+7
2.5E+7
3.0E+7
0% 5% 10% 15% 20%
Co
mp
lex
She
ar M
od
ulu
s, G
* (P
a)
LMS%
y = 2623.3e61.1x R² = 0.9336
1E+0
1E+2
1E+4
1E+6
1E+8
0% 5% 10% 15% 20%
Co
mp
lex
She
ar M
od
ulu
s, G
* (P
a)
LMS%
16
the effects of solvent and air species. It can be found that both virgin and RAP binder
components have molecular weights ranging from 20,000 to 200 Daltons, while those
from RAS binder starting as high as 50,000 Daltons. Therefore, a more reasonable LMS
fraction can be re-defined as the area under the curve with molecular weights higher than
a specific threshold that is defined as large molecule threshold (LMT). Correspondingly,
LMS% can be calculated using Eq. 3-1.
Figure 3-3. GPC Chromatograms Based on Molecular Weights
(3-1)
According to literature review, the only research that has addressed the binder
component fractions based on molecular weight distribution, has divided the curve into
three fractions including polymers (molecular weight greater than 19,000), asphaltenes
(molecular weight from 19,000 to 3,000), and maltenes (molecular weight less than
3000). Since the purpose is to clearly differentiate virgin-recycled binder blend with
different blending levels, the threshold that yields the most significant difference in
17
LMS% should be considered.
Figure 3-4 plots the relation between LMT ranging from 1,000 to 20,000 Dalton
and LMS% differences between RAP/RAS and virgin binders based on the solution in
concentration of 1 mg/mL. Three types of virgin binder, RAP and RAS from different
sources were selected, respectively. The peak of LMS% difference is always found
around 3,000 Dalton. This finding can be related to the above mentioned study that
divided the molecular weight range from 3,000 to 19,000 into asphaltenes fraction, which
is usually considered as the large molecular components fraction for asphalt binder.
Therefore, 3,000 was selected as LMT for calculation of the newly defined LMS%.
(a) RAP-Virgin Binders
18
(b) RAS-Virgin Binders
Figure 3-4 Relation between LMT and LMS% Difference for RAP/RAS-virgin binders
Figure 3-5 presents LMS% values calculated with 3,000 Dalton as LMT. It can be
seen that LMS% values for virgin binders selected in this study range from 10.1% to
20.5%, while those for RAP binders lie in a clearly higher range from 23.8% to 27.1%.
The RAS binders yield an even higher range from 36.3% to 43.5%. The variations among
triplicates are small with maximum coefficient of variation (CV) of 2.34% obtained from
Tennessee RAP B binders, which validates the repeatability of this LMS% determination
method.
19
Figure 3-5. LMS% calculated with LMT of 3,000 Dalton for different binders
3.1.2 Atomic Force Microscopy (AFM)
In our study, the second question of interest is how well the aged binder blends
with the virgin binder? Figure 3-6 shows the two possible scenarios for blending between
the mobile aged binder and virgin binder, where either aged binder partially dissolved
into the virgin binder, or a total blending into a homogeneous “new” material. Therefore,
this question can be answered by investigation of the homogeneity of the blended binder.
20
Figure 3-6. Two scenarios for blending
The two aforementioned possible scenarios in Figure 3-6 cannot be differentiated
if blended binder is dissolved in solution, so the binder homogeneity defined in this
research needs to be studied in solid state without extraction. The direct detection can be
achieved by engaging microscopy, but due to the opacity of bitumen, higher resolution
was not attainable by any traditional microscopic technique [28]. The use of atomic force
microscopy (AFM) to characterize bitumen made it possible to reveal the details of the
material’s microstructural morphology [35].
21
Figure 3-7. Typical topographic profile of one domain on the surface of virgin binder
observed in AFM
The model of atomic force microscope used in this study was the “Multimode-V
Atomic Force Microscope” from Bruker (Santa Barbara, USA), consistent with previous
research [31, 34, 35]. The tapping-mode [36] was chosen because it is ideal for soft
material such as bitumen [36, 37]. During imaging, a cantilever with a sharp tip was
oscillated and scanned across the surface of the sample. Two types of images, topography
(Figure 3-7) and phase-contrast (Figure 3-8), were acquired from a piezoelectric scanner
and post-processed to be readable.
22
Figure 3-8. Illustration of phase lag between set signal (free oscillation) and the tip
response due to sample-tip interaction (adapted from [28, 35])
The cantilevers used in this study were made from Antimony (n) doped Silicon,
with nominal dimensions of 120μm×35μm×3μm. The drive frequency was 330 kHz and
the nominal force constant was 40 N/m. The probe rate was 1.0 Hz and the pixel
resolution was 512×512. Scan sizes were mostly 30×30 μm, with a few were 15×15 μm.
3.2 Laboratory Methods to Determine Bleeding Efficiency
3.2.1 Building “Blending Chart”
A total of 11 points, with RAP/RAS binder content increasing from 0 to 100% in
10% interval, were selected to ensure accuracy. Duplicates were made for each point.
Since only 10 to 15 mg of the sample is required for GPC test, sampling after traditional
mechanical blending may cause high variations thus should be avoided. The virgin binder
and recovered RAP/RAS binder were first dissolved into THF to make solutions in target
concentration (1 mg/mL), then 10 mL solution of the binder blend was produced by
adding solutions of different binders in proportion to a 20 mL scintillation vial (e.g. 8 mL
23
virgin binder and 2 mL RAP/RAS binder solutions were added to produce a solution of
binder blend containing 20% RAP/RAS binder. The solution was shaken in a solution
shaker at high speed for 1 minute for complete dissolution, and then injected through a
0.2 μm filter to filter out the undissolved impurities. An auto-sampler was used to collect
the prepared sample and then placed in the sample holder in EcoSEC GPC. 15 minutes
were required for running one sample.
3.2.1 Aged Binder Mobilization Rate
Figure 3-9 illustrates two possible blending scenarios when recycled asphaltic
material is used, where RAP represents both RAP and RAS hereinafter. The ideal case
would be total mobilization of RAP binder and then full blending with virgin binder,
which generates a homogeneous film coating both the virgin and RAP aggregates. In
reality, however, part of the RAP binder remains inactive and cannot be mobilized during
mixing, thus not contributing to further coating. On the other side, the mobilized RAP
binder may highly interact with virgin binder, which yields a binder blend serving to coat
both the virgin and RAP aggregates.
24
Figure 3-9 Two possible blending scenarios
The mechanical “scrape-off” has been studied as a factor to address RAP binder
mobilization by researchers through “dry blending”, which is a mixing process without
virgin binder addition [1, 38]. However, the authors admitted the impact of the presence
of hot virgin binder cannot be captured by “dry blending”.
The actual blending process in the asphalt plant is complicated since RAP/RAS
and virgin binders cannot be differentiated after mixing. According to Figure 3-9,
however, the binder attached to the virgin aggregates can be evaluated in the
representative of the binder blend of virgin binder and mobilized RAP binder. Therefore,
the mobilized RAP binder contribution in the blend can be analyzed, if virgin aggregates
can be separated after mixing. Assuming the relation between the RAP binder content and
certain specific parameter is known, and this parameter of the binder blend can be
measured through corresponding experiment, the RAP binder content in the blend can be
quantitatively determined.
As illustrated in Figure 3-9, RAP binder content in the blend can be expressed as
Eq. 3-2:
(3-2)
RAP binder mobilization rate, denoted as αM, is defined as
(3-3)
Given percent RAP binder by total mixture as P(b, RAP), percent virgin binder by
total mixture as P(b, Virgin), Eq. 3-2 can be further written as
25
(3-4)
Then αM can be solved out as
(3-5)
Table 3-2 presents an example of αM calculation in a real case. Assuming RAP
Binder (%) Blend is determined in the lab as 20%, αM can be calculated since the rest of the
parameters are all given if one paving job is finalized. For this case, αM is calculated as
75%, which means 75% of the RAP binder can be mobilized during mixing and
contribute to coating the aggregates.
Table 3-2 Example of αM calculation
Parameters Description Calculation Value
RAP (%) Mixture RAP (%) by total mix Given 30%
RAP Binder (%) RAP RAP binder (%) in RAP Given 5%
Pb Optimum binder (%) by total mix Given 6%
P (b, RAP) RAP binder (%) by total mixture [RAP (%) Mixture] ·
[RAP Binder (%) RAP]
1.5%
P (b, Virgin) virgin binder (%) by total mixture Pb − P (b, RAP) 4.5%
RAP Binder (%) Blend RAP binder (%) by binder blend
(after mixing)
Determined in the lab 20%
αM RAP binder mobilization rate Eq. (3.2b) 75%
According to the aforementioned case, the key point of αM calculation lies on
determination of the RAP Binder (%) Blend. A straightforward method to determine RAP
Binder (%) Blend is to build a “blending chart” with one certain parameter and
interpolating the lab-testing result of the sample. As mentioned in “Introduction” section,
a new parameter for blending research derived from GPC testing is to be developed in
this study. Accordingly, this study follows the flow chart presented in Figure 3-10.
26
Figure 3-10 Research flow chart of this study
3.2.3 Staged Extraction Method
Researchers have used a method called staged extraction [1, 39], or progressive
extraction [40], that soaks the asphalt mixture in asphalt solvent for a certain period of
time so that the binder can be extracted layer by layer from the aggregate after mixing.
The previous research has yielded promising results, however, several questions
still need to be answered before it can be widely used. Does the solvent affect the
analysis? Do the light components in the binder tend to be extracted first, which may lead
to a fake “layered structure”? Which is the best or most effective solvent? Is the RAP
binder homogeneous after years’ service? This paper answered these questions by
validating the staged extraction method and used it to further investigate the binder
27
homogeneity of RAP-virgin blend. Since RAS is similar to RAP but stiffer, the same
approach validated on RAP can be extended to RAS. The binder homogeneity research
on RAS-virgin blend could be even more valuable, because air-blown process and
long-term aging of RAS may result in a more difficult blending potential.
3.2.3.1 Solvent Selection
Trichloroethylene (TCE) has traditionally been used for asphalt extraction and
identified by Strategic Highway Research Program (SHRP) as one of the best solvents.
The health and environmental concerns, however, has limited the use of TCE and
attracted asphalt researchers to a less toxic solvent, the combination of toluene and
ethanol (T/E) (85:15 volumetric ratio), in the past few years. The T/E blend,
unfortunately, has still raised potential health concerns [2]. Recently, the need to replace
chlorinated solvents led state agencies to use alternative normal propyl bromide (nPB)
solvents, and it was found that nPB solvents can be used as direct replacements for the
chlorinated solvent[2, 41]. Another solvent, decahydronaphthalene (decalin), was
evaluated as asphalt solvent since it has similar solubility parameter and dissolution
kinetics to toluene at 15 ºC [42].
Most of the staged extraction studies used TCE as the solvent [1, 39, 43, 44], or
the solvent type was not addressed [40]. None of the research, however, mentioned the
potential effect of the solvent on the results. In this paper, the effects of different solvents
on the binder were evaluated in terms of the change in molecular weight distribution
obtained by gel permeation chromatography (GPC). The selected solvents included TCE,
nPB, T/E (85:15 volumetric ratio) and decalin, which are the ones mentioned in the
literature review.
Figure 3-11 presents the GPC results in terms of LMS%. LMS % is defined as the
percentage of large molecular fraction (molecules larger than 3,000 Dalton) over the total
area of the chromatogram generated by GPC [22]. 100 g of one typical PG 64-22 binder
28
was dissolved in solvents, recovered in a vacuum oven at 85ºC for overnight, then subject
to GPC test. It can be found that LMS % values for the binders extracted and recovered
by the solvents selected in this study are approximately the same, indicating that there is
no or limited effect of the solvent on the molecular weight distribution of the binder.
Figure 3-11 Solvent effect on GPC results
According to the selective dissolution results, TCE seems a better solvent that can
be used in staged extraction research. The dissolving rate of each solvent, however,
should also be checked since the more time the asphalt mixture was soaked in the solvent,
the more uncertainties could be caused. The authors of this study also recorded the
extracted sample weight of each layer from the samples mentioned above. Figure 3-15
shows the plot of the change in percentage of accumulative extracted binder weight over
the total binder with increase of dissolving time. It can be seen that TCE dissolved the
asphalt binder faster than the rest of the solvents. nPB was ranked 2nd
, while T/E and
decalin may extract the binder layer with a similar but slow rate. Accordingly, TCE was
revealed by the dissolving rate finding as the most effective solvent that can be used with
staged extraction method. NPB can the considered as the replacement if TCE is not
29
accessible, but may slightly affect the analysis. Therefore, TCE was used as the solvent
for the rest part of the paper.
Figure 3-15 Dissolving rate of different solvents
3.2.3.2 Selective Dissolution
Staged extraction method has been used on the basis of the assumption that all the
components in the asphalt binder can be extracted layer by layer in the same proportion.
However, this assumption has been questioned. Since outer layers extracted from RAP
aggregate were found to be softer than inner layers in plenty of studies [1, 39, 40, 43, 44],
one concern has been raised that lighter maltene fraction of the binder may wash out first,
then the remaining asphaltene fraction will breakdown after successive washes [42]. To
address this concern, the same virgin binder was fractionated into maltene and asphaltene
by soxhlet extraction, with iso-octane as the fractionation solvent (Figure 3-12).
30
Figure 3-12 Fractionation of asphalt through soxhlet extraction method
Figure 3-13 presents the GPC results of the asphaltene and maltene from soxhlet
extraction, as well as the control virgin binder. It can be seen that asphaltene yielded
significantly different chromatogram from corresponding virgin binder, while maltene
followed the same shape but with smaller large molecule fraction. LMS% values can also
be differentiable, with approximately 20, 50 and 11 for control binder, asphaltene and
maltene, respectively. On the basis of this finding, a layered extraction design was
brought up to evaluate if selective dissolution occurred.
31
(a)Chromatograms (b) LMS%
Figure 3-13 GPC results after binder fractionation
Staged extraction by each solvent was conducted on a 500 mg sample of the
same binder. The sample was rolled into the ball-shape prior to the extraction so as to
avoid the potential effect of the sample geometry, and then washed by the solvent for 30
seconds for 7 times. Each washed layer was labeled as layer 1 to 7 outside in, with the
remaining layer designated as layer 8. GPC test was conducted on each layer, and the
results can be found in Figure 3-14. “Total” represents the binder totally dissolved in TCE
and recovered afterwards, serving as the control sample. There was no appreciable
difference in terms of LMS% observed on each layer extracted by TCE and the
corresponding control sample. nPB did not yield much difference either, but the LMS%
values of the first several layers were slightly lower than the layers extracted later and the
control sample, which indicates that slight selective dissolution might occur. The average
of LMS% of the layers extracted by toluene/ethanol (T/E) combination was found to be
approximately equal to the control sample. However, the dissolution of the components
seems unstable. This indicates that selective but non-sequential dissolution might occur
when T/E combination is used as the asphalt solvent. The most significantly selective
dissolution was found on samples extracted by decalin. This finding is consistent with the
binder fractionation results from a previous study [42], and it can be concluded that the
concern over decalin used as an asphalt solvent still stays.
32
Figure 3-14 GPC results after staged extraction by different solvents
3.2.3.3 Binder Homogeneity of RAP and RAS Particles
The homogeneity of the binder covering the RAP and RAS aggregates can affect
the blending process. Prior to using the staged extraction method, the binder homogeneity
of the RAP and RAS materials ready for mixing should be checked. In this study, the
same RAP and a locally available RAS batch were checked for corresponding
homogeneity.
Two identical samples of 50 g for each were selected for each material. Each
sample was soaked in TCE for 1-minute wash of three consecutive times, and then left in
the 4th
beaker filled with TCE for 30 minutes for a complete dissolution. The sample of
each layer was then prepared into standardized sample and tested in GPC.
Figure 3-16 presents the results. There is no difference among the layers that can
be detected. In addition, the LMS% of each layer was found to be equal to that of the
totally extracted control sample. The similar finding can be applied to both RAP and RAS
33
selected in this study. This means the binder coating the RAP or RAS aggregate is
homogeneous after long years’ service, which enables using the staged extraction method
to investigate the recycled and virgin binder blending.
(a) Results of RAP (b) Results of RAS
Figure 3-11 GPC results of four-layer stripping of raw RAP and RAS
3.3 Factors Affecting the Blending Efficiency
3.3.1 Mixing Factors
3.3.1.1 RAP
Materials used in the mixture were Performance Grade (PG) 64-22, RAP of an
unknown source, and a virgin aggregate. The RAP was processed by first barrel rolling
two minutes and then sieved eight minutes, collecting only material retained on the #4
and #8 sieves. The process was for assurance of no agglomerates and minimizing dust.
An extraction and recovery of the processed RAP was performed with a resulting asphalt
content of 3.24%.
The virgin aggregate was sieved and the material retained on the 1/2-inch sieve
was collected. The sizes were chosen based on the ability to readily distinguish the two
materials after mixing with the virgin binder. A total asphalt binder content of 3.00
34
percent was selected with a rap binder replacement of 2.10 percent. The percentages were
selected based upon trial blending and recovery testing. Additionally, based on
preliminary trials, it appeared that the smaller aggregate size of the RAP caused it to
receive priority in coating due to increased surface area. The percentages of RAP, virgin
binder and RAP binder replacement for the mix were selected for assurance of total virgin
binder coating on the smaller RAP aggregates did not saturate the RAP to an unrealistic
level and reduce the margin of the ability to distinguish virgin binder from RAP binder.
The surfactant based WMA additive Evotherm and the wax based WMA additive Sasobit
were additionally studied. Both additives were added to the PG64-22 binder for their
respective mixtures.
The virgin binder was heated to mixing temperatures on the basis of the test
matrix and the virgin aggregate was superheated to 10C beyond the mixing temperature.
The RAP was not heated so that any blending was induced by the superheated aggregates,
similar to the processes that occur in the asphalt plant.
In efforts to ensure that the virgin binder coated the coarse aggregate the
aggregate and binder were mixed for one minute prior to the addition of RAP. The RAP
was then added and allowed to mix for the duration of mixing time provided in the
experimental matrix (Table 3-3). This step was important because the RAP aggregates are
smaller in size. The virgin binder was found to have a tendency to be attracted to the
smaller aggregates, and coats them first. This caused an irregular amount of virgin asphalt
to coat the RAP without coating the larger, virgin aggregates. A Hobart Mixer model
A-120 with wire whisk was used for mixing.
Table 3-3. Experimental matrix of mixing scenarios
Mixture Time (s) Temperature (C)
1 30 160
2 60 160
3 105 160
4 150 160
5 300 160
35
The virgin binder was heated to mixing temperatures on the basis of the test
matrix and the virgin aggregate was superheated to 10C beyond the mixing temperature.
The RAP was not heated so that any blending was induced by the superheated aggregates,
similar to the processes that occur in the asphalt plant.
In efforts to ensure that the virgin binder coated the coarse aggregate the
aggregate and binder were mixed for one minute prior to the addition of RAP. The RAP
was then added and allowed to mix for the duration of mixing time provided in the
experimental matrix (Table 3-4).
Table 3-4. Experimental matrix of mixing scenarios
Mixture Time (s) Temperature (C) Additive
1 105 130 -
2 105 160 -
3 105 180 -
4 105 130 Evotherm
5 105 130 Sasobit
At the conclusion of mixing, the large, “coarse” (virgin) and small, “fine” (RAP)
aggregates were separated. Upon separation, the binder was recovered from the mixtures
using AASHTO T164 “Standard Method of Test for Quantitative Extraction of Asphalt
Binder from Hot-Mix Asphalt (HMA)”. The solvent used for the extraction and recovery
was n-propyl bromide. The recovered binders were then subject to GPC and DSR tests
for evaluating the blending efficiency.
3.3.1.2 RAS
Aggregates of three different sizes that could be visually detected and separated
were designed to be mixed with RAS and virgin binder to make mixtures (Figure 3-17).
The small aggregates were pre-blended with RAS as mixing carriers that completely
covered the RAS before blending, in order to avoid further binder loss to the mixing
bucket. The change in percentages of LMS among the extracted and recovered binder
36
from the different types of aggregates would give a better understanding of the blending
efficiency. In addition, RAS content and mixing time were considered as two parameters
that affect the binder blending efficiency and would be taken into consideration in this
study. Figure 3-18 shows the design parameters for the mixtures studied in this research.
Altogether, seven mixtures were made in the lab and twenty-one GPC samples were
evaluated (Table 3-5). Each of these mixtures is referred to as mixing binder.
Table 3-5 Differentiation between mixtures 1 through 7
Time Temperature (C) RAS Content (%)
Mixture 1 2 min 170 2.5
Mixture 2 2 min 170 5
Mixture 3 2 min 170 7.5
Mixture 4 2 min 170 10
Mixture 5 30 s 170 5
Mixture 6 1 min 170 5
Mixture 7 3 min 170 5
Figure 3-17 Aggregates and RAS used in this study (size distribution in Table 3-5)
37
Figure 3-18 Design parameters for mixture blending
The RAS used in this study came from tear-offs, which was plant screened and
passed No. 4 sieve and had an asphalt content of 19%. The virgin binder is a typical PG
64-22 binder commonly used in the United States. The aggregates utilized in this study
included natural sand (referred as small) that completely passed No. 4 sieve, intermediate
limestone (referred as medium) that totally passed 19 mm (¾-in.) sieve but retained on
No. 4 sieve, and coarse limestone (referred as large) that retained on 19 mm (¾-in.) sieve.
Table 3-6 summarized the size distribution for materials used in this study. Gradation was
ignored but each component of the mixture was given specific portion by weight of total
mix to make it close to a well-graded mixture (Table 3-7). Among all the mixtures, there
were 10% large aggregates, 40% medium aggregates, and 36–42% small aggregates that
38
vary with the change of RAS content. The 5.5% optimum asphalt content was selected
for all the mixtures.
Table 3-6 The size distribution of the materials
Materials Size Distribution
RAS Pass No.4 Sieve
Small Aggregates Pass No.4 Sieve
Medium Aggregates Pass 3/4 in Sieve, Retain on No.4 Sieve
Large Aggregates Retain on 3/4 in Sieve
Table 3-7 Mix design for mixtures 1 through 7 (by weight of the total mixture)
RAS
(%)
Large
Aggregate
(%)
Medium
Aggregate
(%)
Small
Aggregate
(%)
Virgin
Binder
(%)
Total
(%)
Mixture 1 2.5 10 40 42.475 5.025 100
Mixture 2 5 10 40 40.45 4.55 100
Mixture 3 7.5 10 40 38.425 4.075 100
Mixture 4 10 10 40 36.4 3.6 100
Mixture 5 5 10 40 40.45 4.55 100
Mixture 6 5 10 40 40.45 4.55 100
Mixture 7 5 10 40 40.45 4.55 100
The RAS was extracted and recovered according to the centrifuge-rotavapor
recovery method [45]. The recovered RAS binder was heated up to 240C at which the
binder was flowable enough to be blended, while the virgin binder was heated up to
154C. Then the two binders were blended by a high-performance mixing gun at 180C
39
with a total weight of 100 g for 3 min. The newly blended binder was made into DSR
standard samples with 8 mm diameter and GPC samples that weighed between 15 to 20
mg. The results were obtained based on the average of two duplicates.
The aggregates were heated to 180C, 10C higher than target mixing temperature
of 170C for more than 2 h before mixing. RAS samples were heated to 110C for 2 h to
avoid further aging. Virgin binder was heated to 170C for 2 h. The small aggregates
were pre-blended with RAS samples as RAS carriers 20 min before mixing with the
larger aggregates to avoid further binder loss during mixing. The three different types of
aggregates were separated immediately after mixing was complete, among which a
representative (around 50 g) of each aggregate was selected to be extracted, recovered,
and tested by GPC. The results were obtained based on the average of two duplicates.
3.3.2 Effect of Aged Binder Content
In this study, gravel with round and smooth surface passing 9.5 mm (3/8 in.) sieve
and retained on 4.75 mm sieve (No. 4) was selected and added as part of the virgin
aggregates in order for better separation. Figure 3-19 shows the round gravel separated
before and after mixing.
Figure 3-19. Round-shaped gravel before and after mixing
40
A total of 10 mixtures with 2,000 g for each batch, covering RAP percentage up to
80% and RAS up to 10% (Table 3-8), were prepared in the lab at 165 oC with the mixing
time set to 2 min. 100 g of round-shaped gravel was added as part of the virgin aggregate
for each mix. The properties of RAP, RAS and virgin aggregate used in this study can be
found in Table 3-9. The mix design recommended by the asphalt plant providing the
materials was followed, with consideration of the total contribution of the recycled
binder. The gradation and optimum asphalt content (5.7%) were hold constant for all the
mixtures.
Table 3-8. Usage of recycled materials selected in this study
Recycled Material Percentage by Weight of the Total Mix
RAP 10%, 20%, 30%, 40%, 50% and 80%
RAS 2.5%, 5%, 7.5% and 10%
Table 3-9. Properties of RAP, RAS and virgin aggregate
Properties Mixture Virgin Agg. RAP RAS
AC Content (%) 5.7 - 4.76 20.85
Gradation
Sieve (in.) Sieve (mm) Passing (%)
5/8 16 100.0 100 100 100
1/2 12.5 96.0 92 100 100
3/8 9.5 85.5 71 92.4 100
#4 4.75 59.3 23 61.5 99.2
#8 2.36 41.3 15 44.5 97.4
#30 0.6 19.3 9 26.7 60.2
#50 0.3 11.0 6 18.3 53
#100 0.15 6.6 4 13 44.9
#200 0.075 4.4 2.5 8.7 34.6
After mixing, the round-shaped gravel was picked and eluted by n-propyl bromide
(nPB) for 30 min in order for complete binder extraction. Then the binder was recovered
in a 20 mL vial using a water bath at 70oC in less than 15 min until solvent was
non-visible, and then left in a vacuum oven at 85oC overnight for complete removal of
the solvent. The relatively lower temperatures and vacuum oven were selected to reduce
41
the potential aging effect. The effects of mixing, nPB extraction and recovery were also
checked and found to be limited or none on LMS%. Once the binder sample was ready,
the GPC test was carried out following the same procedure described in “blending chart”
section.
3.3.3 Effect of WMA technologies
A gap-gradation was used to design mixtures in this study in order to easily
distinguish the virgin aggregates from RAP particles after mixing. Two models,
coarse-virgin with medium-RAP and medium-virgin with fine-RAP, were used to design
RAP mixes. The medium-virgin with fine-RAS model was used for RAS mix only, since
most of the processed RAS consists of fine particles passing No.4 sieve. A No.57
limestone batch was selected as the coarse virgin aggregate and the materials retained on
1/2 in. sieve were collected only. The medium-virgin aggregates were collected from a
gravel batch retained on No.4 sieve. A locally available batch of RAP passing 5/8 in. but
retained on No. 8 sieves was used as medium-RAP, while the same RAP and a local RAS
batch passing No.8 were used as fine-RAP and fine RAS. Two types of models were
selected for RAP mix in order for consideration of size effect. Three mixtures were
design to cover 50% RAP and 10% RAS. Table 3-10 presents the properties for each
batch of materials, including gradation, asphalt content and average film thickness
calculated following the method proposed by Asphalt Institute (1993) [3, 38]. A typical
PG 64-22 binder was selected as the virgin binder.
Table 3-10 Properties of the materials used in this study
Properties
Virgin
Agg.
(Coarse)
Virgin
Agg.
(Medium)
RAP
(Medium)
RAP
(Fine)
RAS
(Fine)
AC (%) - - 3.24 9.75 21.28
Film Thickness (μm) - - -* 6.17 10.42
42
Gradation
Sieve
(in.)
Sieve
(mm) Passing (%)
1 25.4 100 100 100 100 100
3/4 19 64.6 100 100 100 100
5/8 16 43.1 100 100 100 100
1/2 12.7 0 89.6 100 100 100
3/8 9.5 0 62.3 86.3 100 100
#4 4.75 0 0 30.6 100 100
#8 2.36 0 0 0 100 100
#16 1.18 0 0 0 77.8 83.1
#30 0.6 0 0 0 60 61.8
#50 0.3 0 0 0 41.1 54.4
#100 0.15 0 0 0 29.2 46.1
#200 0.075 0 0 0 19.6 35.5
*Only the film thickness of fine particles can be calculated using this method.
The non-foaming additives selected are presented in Table 3-11.
Table 3-11 Properties of WMA additives
Type Products Code Description
Evotherm
Related
Evotherm M1 Ev-M1
Chemical packages, liquid, fatty amine
derivatives, 0.25 to 0.75% by weight of
binder
EvoFLEX CA Ev-CA Chemical packages, liquid, fatty acid
derivatives, 1 to 5% by weight of binder
Rediset
Rediset
LQ-1102C Re1102
Chemical packages, liquid, proprietary
alkoxylated fatty polyamines, proprietary
polyamine, Glycol. 0.5% to 1% by weight
asphalt cement
Rediset LQ-1106 Re1106 Chemical packages, liquid, surfactant blend,
0.5% to 1% by weight asphalt cement
Cecabase Cecabase RT 945 Ce
Chemical packages, liquid, fatty acid
amines, 0.3% to 0.5% by weight asphalt
cement
Sasobit
Sasobit Sa Fischer-Tropsch wax, solid, solid saturated
hydrocarbons, 1% to 1.5% by weight of the
binder Sasobit LM Sa-LM
Sasobit Sa-570 Sasol wax Slack wax blend, solid,
43
GTRM570 petroleum hydrocarbon, 1% to 1.5% by
weight of the binder Sasobit
GTRM850 Sa-850
Since the aggregate batch is gap-graded, the mix design is conducted according to
trial and error in order to obtain mixtures with good coating (Table 3-12).
Table 3-12 Mix design
Mixture RAP/
RAS
(%)
Total
Weight
(g)
RAP/
RAS
(g)
RAP/RAS
AC
(%)
Total
AC
(%)
Virgin
Binder
(g)
Virgin
Agg.
(g)
Coarse-Virgin
Medium-RAP
50 2000 1000 3.24 3.00 27.6 972.4
Medium-Virgin
Fine-RAP
50 2000 1000 9.75 6.66 35.7 964.3
Coarse-Virgin
Medium-RAS
10 2000 200 21.28 3.29 23.2 1776.8
The foamed binder was produced at 135oC with Wirtgen model WLB 10S in the
materials lab at Virginia Center for Transportation Innovation & Research (VCTIR), with
water content of 2.5% and 5%, respectively. The non-foaming additives were added and
blended with 100 g virgin binder in a metal container (76.2 mm (3 in.) diameter × 55.9
mm (2.2 in.) height) with a high shear-rate mixing gun at 135oC for 3 min in materials lab
at University of Tennessee, Knoxville (UTK). The additive dosages selected in this study
were generally based on the recommendations by the suppliers. For some additives, the
increased dosage was also used by the authors to extend the testing range.
The mechanism of most WMA technologies is to reduce the viscosity of asphalt
binder so as to be workable at lowered temperature. Thus, viscosity of foamed binder and
binder blended with selected additives was determined, respectively. Brookfield
rotational viscometer was used in accordance with AASHTO T316 to test the viscosity of
the control and WMA binders at 135oC, and one control binder at 165
oC.
44
The binder produced with different WMA technologies were mixed with virgin
aggregates and virgin binder at 135oC following the mix design (Table 3-12). The control
hot mix was produced at 135oC and 165
oC, respectively. It should be noted that foamed
WMA and corresponding control HMA were mixed with Troxler PMW high energy
asphalt mixer at VCTIR, while non-foaming WMA mixes and control mixes were mixed
at UTK with a Hobart Mixer model A-120 with wire whisk recommended by Asphalt
Institute for laboratory mixing. Upon completion of mixing, the materials were separated
into coarse and medium particles, or medium and fine particles, respectively (Figure
3-20). The separated particles were eluted with n-propyl bromide (nPB) for a complete
extraction, then recovered in a 20 mL vial with a water bath of 70oC in less than 15 min
until solvent was non-visible.
Figure 3-20 Separated particles after mixing
The recovered binder was made into standardized sample and subjected to gel
permeation chromatography (GPC). The whole experimental plan is illustrated in Figure
45
3-20.
Figure 3-20 Flow chart of the experimental design
3.3.3 Stage Extraction Method
3.3.3.1 Effect of Extraction Time
Two mixtures, one containing 50% RAP and the other one containing 10% RAS,
were prepared to fulfill the current tendency of incorporating high amount of the recycled
materials. During the asphalt mixture production, the virgin binder may fully blend with
the activated recycled binder and re-coat the virgin aggregates and recycled aggregates
with un-mobilized old binder still remaining. Therefore, the blending occurring on
RAP/RAS aggregates are of higher significance. In order to visually and easily
distinguish the RAP/RAS aggregates from the virgin ones, a gap-gradation was used to
46
design the two mixtures.
The RAP and RAS used in this study were sieved prior to mixing, and only the
materials passing No. 8 (2.36 mm) sieve were collected. The virgin aggregates were from
a gravel batch that is locally available for surface mixture, and sieved for collecting the
part retaining on sieve No. 4 (4.75 mm). The aggregate gradation of the mixture, as well
as the job mix formula (JMF) from the same asphalt plant that provided all the materials,
can be seen in Figure 3-22. The asphalt content of the recycled materials and the mix
design are arranged in Table 5.1. 5.5% was provided by the asphalt plant as the optimum
asphalt content (AC) for the mixture of which the JMF was also provided in Figure 3-21.
The same optimum AC was selected for the 50% RAP mixture since its gradation was
similar. 3.5% was selected for 10% RAS mixture based on trial and error, in order for a
mixture with good coating by visual judgment.
Figure 3-21 Aggregate gradation
Table 3-13 Mix design of the mixtures used in this study
Mixture Total RAP/RAS RAP/RAS Optimum Virgin Virgin
47
Weight
(g)
(g) AC
(%)
AC
(%)
Binder
(g)
Aggregate
(g)
50% RAP 2000 1000 5.71 5.5 52.9 947.1
10% RAS 2000 200 20.85 3.5 28.3 1771.7
The virgin aggregates were pre-heated at 175ºC two hours prior to the mixing,
while the RAP and RAS were heated at 110ºC 30 minutes earlier to gain some
workability. Note that the heating for RAP/RAS should be limited to maximum of 30
minutes to avoid further aging [2]. The virgin binder was heated at 165oC for a minimum
of one hour with the cap tight. A Hobart Mixer model A-120 with wire whisk
recommended by Asphalt Institute for laboratory mixing was used to make mixtures in
this study. A 2-minute mixing was conducted to ensure a better coating. Upon completion
of mixing, the coarse virgin and fine RAP/RAS aggregates were manually separated.
Figure 3-22 presents fine RAP aggregates and coarse virgin aggregates separated after
mixing.
Figure 3-22 Separated fine RAP and coarse virgin aggregates
The equal time interval was commonly used for staged extraction by researchers.
30 seconds, 1 minute [39] and 3 minutes [1] were tried in different studies. However,
48
based on the plenty of trials, the authors of this study found the binder layers extracted
with equal-time extraction may not accurately represent the actual layers. According to
trial and error, a new procedure, named as “Step-Extraction”, was used to strip the binder
layers from RAP/RAS aggregates. The step-extraction included six TCE washes that
yielded six layers for analysis. The time was 5, 10, 15, 20 and 120 seconds for the first 5
washes, respectively. The remaining binder was soaked in TCE for 30 minutes for
complete dissolution, generating the 6th
layer. Figure 3-23 shows the percent weight of
each layer based on a 1-minute equal extraction and re-designed step-extraction. It can be
found that the percent weight of each layer by step-extraction can reach an approximately
equal distribution. For comparison purpose, the 1-minute extraction was also conducted.
A sample of 10 g was used so as to fit a 50 mL beaker. Upon completion of stripping the
binder from RAP/RAS aggregates, each layer was made into standardized GPC sample as
mentioned above.
Figure 3-23 Layer weight distribution caused by different extraction methods
3.3.3.2 Diffusion Studies
A locally available PG 64-22 binder was selected as virgin binder in this study.
49
One gravel batch specified for surface mixture was used as virgin aggregates. One locally
available RAP and one RAS were selected as the recycled materials. As illustrated in
Figure 3-25, the binder distribution on virgin aggregates and RAP/RAS aggregates may
be highly different, so a better analysis can be achieved by evaluating the blending status
on virgin and RAP/RAS aggregates, respectively. Thus, a gap-gradation was selected in
this study in order to easily distinguish the RAP/RAS aggregates from the virgin ones
upon completion of mixing. The virgin aggregates retained on No. 4 (4.75 mm) sieve and
the recycled materials passing No. 8 (2.36 mm) sieve were collected for use in this study.
The basic information of the raw and sieved materials can be found in Table 3-14.
Table 3-14 Properties of materials selected in this study
Properties Virgin
Agg.
Virgin
Agg.
(+ No.4)
RAP RAP
(- No.8)
RAS RAS
(- No.8)
AC (%) - - 4.76 9.75 20.85 21.28
Gradation Sieve
(in.)
Sieve
(mm)
Passing (%)
5/8 16 100 100 100 100 100 100
1/2 12.5 92 89.6 100 100 100 100
3/8 9.5 71 62.3 92.4 100 100 100
#4 4.75 23 0 61.5 100 99.2 100
#8 2.36 15 0 44.5 100 97.4 100
#16 1.18 12 0 34.6 77.8 80.9 83.1
#30 0.6 9 0 26.7 60 60.2 61.8
#50 0.3 6 0 18.3 41.1 53 54.4
#100 0.15 4 0 13 29.2 44.9 46.1
#200 0.075 2.5 0 8.7 19.6 34.6 35.5
Two mixtures, one containing 50% RAP and one with 10% RAS, were prepared
to meet the current tendency of using high amount of recycled materials. Mix design was
adjusted to pursue mixtures with good coating based on visual judgment. The mix design
parameters and other useful properties of mixtures are presented in Table 3-15.
50
Table 3-15 Mix design and other properties of mixtures
Mix Total
Weight
(g)
RAP/RAS
(g)
RAP/RAS
AC
(%)
Mix
AC
(%)
Virgin
Binder
(g)
Virgin
Aggregate
(g)
50%
RAP
2000 1000 9.75 6.66 35.7 964.3
10%
RAS
2000 200 21.28 3.29 23.2 1776.8
Mixing was conducted at 165ºC for 2 minutes with a Hobart Mixer model A-120
with wire whisk recommended by Asphalt Institute for laboratory mixing. Upon
completion, the coarse virgin and fine RAP/RAS aggregates were separated for staged
extraction, respectively. Figure 3-24 presents the separated coarse virgin and fine RAP
particles for 50% RAP mix. Trichloroethylene (TCE) was used as the extraction solvent
since it was found to be the most effective binder solvent used for staged extraction
analysis [42]. A step-extraction was selected to generate binder layers with similar
thicknesses based on trial and error (Table 3-16). A total of four layers and six layers were
obtained from coarse virgin aggregates and fine RAP/RAS aggregates, respectively.
Samples of 25 g and 10 g were used for extraction of fine particles and coarse particles,
respectively, to fit the dimension of a 50 mL beaker. The same aforementioned GPC
testing procedures were conducted on each extracted binder layer.
51
Figure 3-24 Separated coarse and fine particles for 50% RAP mixture
Table 3-16 Step-extraction procedure
Layer No. 1 2 3 4 5 6
Wash Time Coarse 10 s 20 s 30 s 30 minutes - -
Fine 5 s 10 s 15 s 20 s 2 minutes 30 minutes
Figure 3-25 presents the temperature change of hot and warm mix from
production to placement. Since diffusion is highly dependent on temperature [46-48],
diffusion may mostly occur at mixture storage period at higher temperature after the
mixture is produced. In order to evaluate the diffusion at this stage, the fine particles
obtained in “blending” study were left in the vacuum oven for 1 hr at target temperature
for HMA and WMA respectively, and then subjected to staged extraction and GPC
testing. The vacuum oven was used to avoid the effect of aging. The experimental design
was organized in Table 3-17.
52
Figure 3-25 Temperature profile for mix production, storage, transportation and
placement (adapted from [48])
Table 3-17 Experimental plan for diffusion study
ID Materials Treatment Method
1 Coarse particles from 50% RAP mix 125oC for 1 hr 4-layer staged extraction
2 Fine particles from 50% RAP mix 125oC for 1 hr 6-layer staged extraction
3 Fine particles from 50% RAP mix 155oC for 1 hr 6-layer staged extraction
4 Coarse particles from 10% RAS mix 125oC for 1 hr 4-layer staged extraction
5 Coarse particles from 10% RAS mix 155oC for 1 hr 4-layer staged extraction
6 Fine particles from 10% RAS mix 125oC for 1 hr 6-layer staged extraction
7 Fine particles from 10% RAS mix 155oC for 1 hr 6-layer staged extraction
53
CHAPTER 4 LABORATIORY TEST RESULTS AND
DISCUSSION
4.1 Blending Chart
Figure 4-1 presents the “blending charts” generated by blending one typical PG
64-22 binder with recycles asphalt pavement (RAP) and/or recycled asphalt shingle
(RAS) binder. All three materials are typical materials and locally available in Tennessee.
Linear correlations between recycled binder content and LMS% are revealed on both
virgin-RAP blend and virgin-RAS blend with high “R2” values (both over 0.98).
Accordingly, the following equations can be derived, for determination of RAP Binder
(%) Blend [Eq. (4-1)] and RAS Binder (%) Blend [Eq. (4-2)], respectively.
(a) RAP-virgin blend
(b) RAS-virgin blend
Figure 4-1. “Blending Chart” generated with LMS%
(4-1)
(4-2)
54
It should also be noted that the proposed idea of using LMS% to build the
“blending charts” between recycled and virgin binders can be extended to other
multi-binder involved studies. Its further application is not discussed in this paper, but
can be valuable for future research.
4.2 Factors Affecting the Blending Efficiency
4.2.1 Effect of Mixing Factors
4.2.1.1 RAP
The recovered binders from the coarse and fine aggregates were tested for their
rheological properties. In a poor mixing condition the fine (RAP) aggregate is expected to
exhibit stiff rheological properties due to the presence of the RAP binder, while the coarse
aggregate is expected to exhibit soft rheological properties since it was mixed with virgin
binder prior to RAP inclusion.
55
Figure 4-2. Master curve of the Complex Modulus (G*) for changing mix times
The master curve data (Figure 4-2) shows that there is a large gap between the
virgin binder and RAP binder curves. In the case of the virgin binder curve, the binder
was recovered from a coarse mixture that contained no RAP. At 30, 60, and 105 seconds,
there is no noticeable differentiation between the coarse master curves. However, when
the mixing time reaches 150 seconds there is an evident increase in the complex modulus
at the lower frequencies. This indicates that there is an increase in the presence of RAP
binder in the binder recovered from the coarse aggregate.
A ratio for the LMS% of the coarse aggregate to the fine aggregate was
considered for an estimation of the efficiency of blending. The RAP binder exhibits a
significantly higher LMS% than virgin binder. Thus, if RAP binder blends with the virgin
binder on the coarse, virgin aggregate, the LMS% should increase. Likewise, as the virgin
56
binder blends or diffuses into the RAP binder on the fine aggregate, the LMS of the fine
aggregate should decrease. If the LMS% for the binder recovered from the coarse
aggregate, which has been increasing in the number of large molecules present, is
equivalent to that of the binder recovered from the fine aggregate which is decreasing in
large molecules as the virgin binder blends and/or diffuses into the RAP binder, then a
complete blend will have been achieved. The blend ratio is computed as follows:
(4-3)
At a typical mixing temperature of 160C the times were increased from 30
seconds to 150 seconds. The blending ratio gradually increased from just over 55% at 30
seconds of blending to nearly 80% after 150 seconds. . This data was plotted and
regressed as shown in Figure 5, to estimate the total time, assuming linearity, to achieve a
100% blend. A linear regression was performed on the data, which yielded the following
equation for the estimation of the blending percentage with respect to time:
(4-4)
The basis of the regression equation, it would take over 5 minutes to achieve a
100% blend. That is not a realistic time frame for mixing in a practical field application.
A 5 minute mixing time was attempted, as shown in Figure 4-3.
57
Figure 4-3. Regression of 30, 60, 105, and 150 second blending ratio results to calculate
100% blend ratio. Blending ratio at 300 seconds also presented.
An important finding as noted in Figure 4-3 is that doubling the mix time beyond
150 seconds to 300 seconds (indicated by square in Figure 4-3) only achieved a one
percent increase in blend ratio from 77% to 78%. In this mixing scenario it can be
concluded that an increase in mixing time is not helpful in creating an increase in blend
ratio, and the trend is certainly not linear beyond 150 seconds.
58
Figure 4-4. Master curve of the Complex Modulus (G*) for changing temperatures
The master curve shown in Figure 4-4 reveals the influence of mixing temperature
on the blending of the binders. The coarse aggregate clearly increases in complex
modulus as the mixing temperature increases from 130C to 160C and from 160C to
180C. Furthermore the fine aggregate master curves never have any clear separation,
indicating that the fine aggregate, originally coated in RAP binder, is not further oxidized
even at higher temperatures. This is further supported when examining Figure 4-5 which
shows the average LMS% and standard deviation for both the recovered fine and coarse
binders.
The data presented in Figure 4-6 shows that there is an increase in blend ratio as
the mixing temperature increases. This is to be expected, however there it should be noted
that since there is no increase in the LMS% for the fine aggregate, it is clear that the
binder on the coarse aggregate is indeed receiving some of the effective RAP binder. This
59
is likely because the hot virgin aggregates and the hot virgin binder can assist in melting
the RAP binder, allowing for further diffusion of the two binders into one another.
Figure 4-5. Course and fine LMS% for mixing times at varying temperatures at a 105
second mix time
60
Figure 4-6. Blend ratio for varying mix temperature at a 105 second mix time
The evaluation of WMA additives compared to a control mixture yielded
interesting results. In both cases, the WMA additive showed to increase the blend ratio.
However, as seen both by the master curve in Figure 4-7 and the blend ratio in Figure 4-8
the Evotherm WMA additive induced better blending than either the Sasobit mixture or
the control mixture. The Evotherm mixture had a blend ratio equivalent to that of the
160C, 150 second mixture even though it was mixed for less time and at a lower
temperature. This indicates that Evotherm actually does assist in the blending of RAP and
virgin binders. The researchers did notice that Sasobit seemed to increase the workability
and more easily coated the RAP when the coarse aggregates with virgin binder were
mixed together. Further investigation into the use of Sasobit at a longer mix time (150
seconds) or a higher temperature (160C) may be worth investigating to see if these two
variables play a significant impact in Sasobit’s ability to increase RAP blending
efficiency.
61
Figure 4-7. Master curve of the complex modulus (G*) for different WMA additives and
a control mixture
62
Figure 4-8. Blend ratio for the 130
oC control and WMA additives at 105 seconds of
mixing
4.2.1.2 RAS
Binders of mixtures containing 5% RAS were selected to evaluate the effect of
mixing time. It can be seen from Figure 4-9 that both the LMS% of binders on medium
and large aggregates expressed upward trends with the increase in mixing time. Unlike
binders on small aggregates, the medium and large ones had not been pre-blended with
RAS prior to laboratory mixing. This means the increase in LMS levels of these two
aggregates reflects an increase in blending with the RAS binder. The larger the
percentage of LMS, the more RAS binder the carriers contributed to the blending. The
small aggregates, on the contrary, didn’t show any apparent trend with the change of
mixing time. As the RAS carrier, small aggregates hold a large amount of RAS binder,
which made it difficult to detect the percentage change in LMS, even if the change in
blending efficiency did occur.
63
Figure 4-9 The relation between mixing time and percentage of LMS
The effect of aggregate size was evaluated using GPC to investigate the blending
efficiency of the RAS binder within a mixture. Figure 4-10 represents seven mixtures,
which have different RAS contents or mixing times. The differences between mixtures 1
through 7 were previously defined in Table 3-6 and 3-7. Though each mixing case was
different, the small carrier aggregates always had a higher LMS content than that of the
medium and large aggregates. Furthermore, the medium and large aggregates always
maintained nearly the same percent LMS which indicates that they had a similar binder
molecular weight distribution, which correlates to G* and further performance. The
similar LMS% result for the Medium and Large aggregates indicates that the size of the
virgin aggregate does not alter the blending efficiency.
The difference in percentage of LMS for the larger aggregates (large and medium)
as compared to the carrier aggregates (small) is an important finding. The increased RAS
content shown in Mixes 1 through 4 resulted in an increase in percentage of LMS in both
medium and large aggregates in all mixes, indicating that the RAS binder is in fact
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
4.5%
0 0.5 1 1.5 2 2.5 3 3.5
LM
S%
Mixing time (min)
Medium Aggregate Large Aggregate Small Aggregate
64
blending with the virgin binder when the mixing process is occurring. The difference of
percentage of LMS between small carriers and larger aggregates indicates that a complete
blend is not occurring between the small carrier aggregate binder and the larger aggregate
binder. Thus it can be concluded that for this mixing case only partial blending
commences.
Figure 4-10 The comparison of small “carrier” aggregates to medium and large
aggregates
The small carrier aggregate binder contained a higher LMS and thus was believed
to have a higher RAS content than the larger aggregates (large and medium). When
evaluating the percentage of LMS for the larger aggregates and comparing it to the small
carrier aggregates an interesting conclusion could be reached. Based on the limited data
provided by the large and medium aggregate tests, the ratio of percentages of LMS of
larger aggregates to small aggregates increased up to 5% RAS and then decreased, as
shown in Figures 4-11 and 4-12. The increase in the ratio indicates that blending is
occurring because the larger aggregate binder LMS percentage is moving closer to that of
0%
1%
2%
3%
4%
5%
6%
7%
0 1 2 3 4 5 6 7 8
LMS%
Mixture Number
Large Aggregate Medium Aggregate Small Aggregate (Carrier)
65
the small carrier aggregate binder. This finding is interesting when compared to past
literature which states that a maximum recommended RAS content is 5% [49].
Figure 4-11 Relationship between the medium aggregate LMS and carrier (small
aggregate) LMS when compared to RAS content
0%
10%
20%
30%
40%
50%
60%
70%
80%
2 4 6 8 10
Med
ium
LM
S/C
arr
ier
LM
S
RAS Content (%)
0%
10%
20%
30%
40%
50%
60%
70%
80%
2 4 6 8 10
Larg
e L
MS
/Carr
ier
LM
S
RAS Content (%)
66
Figure 4-12 Relationship between the large aggregate LMS and carrier (small aggregate)
LMS when compared to RAS content
4.2.2 Effect of Aged Binder Content
Figure 4-13 presents the results of LMS% obtained from GPC. It can be seen, for
both RAP and RAS, that LMS% of the binder blend increases with the increase of
RAP/RAS content in the mixture.
(a) LMS% for RAP mixes (b) LMS% for RAS mixes
Figure 4-13. LMS% results
Figure 4-14 presents the RAP/RAS Binder (%) Blend values. As expected, the trend
is consistent with that of LMS%.The increase of recycled binder content in the blend can
be attributed to the increase of potentially workable RAP binder in the mixture and less
addition of virgin binder, when RAP/RAS percentage is increased.
67
(a) RAP Binder (%) Blend (b) RAS Binder (%) Blend
Figure 4-14. LMS% results
Figure 4-15 shows recycled binder mobilization rates for the mixtures selected in
this study. It is found in Figure 4-15 (a) that the mobilization rate decreases if more RAP
is added, which indicates lower ratio of the available RAP binder by total will be incurred
with increase of RAP addition. For low RAP mixtures (10% and 20%), the mobilization
rates are fairly close to 100%, which indicates an approximately complete mobilization of
the RAP binder during mixing. However, the mobilization rate drops from 73% to 24%
with RAP percentage varying from 30% up to 80%. This finding may lead to an
assumption that the fatigue and cracking resistance of HMA containing high RAP (30%
or over) is reduced not just because the high stiffness of the recycled binder, but also due
to its lower mobilization rate that may cause an under-asphalt mixture or heterogeneous
blending. This assumption should be validated in future research.
Unlike RAP, the mobilization rates of RAS mixtures are comparatively low, even
at low RAS content. This can be explained by the highly aged nature of RAS binder,
which limits the RAS binder from being mobilized at normal mixing temperature. It can
also be seen that the mobilization rates of low RAS mixtures (2.5% and 5%) stay around
50% to 60%, which are close to 66.7%, an engineering estimation that has been used by
several state agencies as the RAS binder contribution factor. The mobilization rate for
10% RAS mixture can be as low as 36%, which may also account for the restriction of
maximum usage of RAS to 3% to 5% in several states. The interesting finding is that the
68
optimum mobilization rate is found on 5% RAS mixture. This may be explained by the
assumption that RAS particles tend to be coated by virgin binder at a low RAS
percentage (2.5%), thus could be limited from providing binder to coat virgin aggregate.
Meanwhile, RAS particles tend to agglomerate with increased percentage (over 5%) thus
leading to less exposure to the binder blend. In addition, the 5% optimum mobilization
rate is consistent with the blending ratio results addressed in a previous study by the
authors [22].
(a) Mobilization rates for RAP mixes (b) Mobilization rates for RAS mixes
Figure 4-15. Calculated mobilization rates for mixes selected
4.2.3 Effect of WMA technology
The viscosity of the asphalt binder is used to reveal its flow characteristics to
ensure that the binder can be pumped and handled on site and also to determine the
mixing and compacting temperature of the mixture [50].
During viscosity testing for foamed asphalt, it was noticed that the reading was
not stable even after 30 minutes, thus, the results cannot be used. This can be explained
by the water trapped in the binder tending to evaporate at testing temperature of 135oC,
so the whole binder system was not in equilibrium. However, the workability of WMA
69
produced with foamed asphalt was fairly good based on visual judgment.
Figure 4-16 presents the results of binder blended with additives and tested at
135oC as well as control binder tested at 135
oC and 165
oC. As expected, the viscosity
values of WMA binders slightly decreased with addition of additives, as compared to
control binder tested at the same temperature (135oC). This indicates that the WMA
additives selected in this study can help reduce the mixing and compaction temperatures.
However, it can also be found that all the viscosity values of WMA binders are
significantly higher than that of control binder tested at regular mixing temperature for
hot mix (165oC). This means the WMA produced by adding additives may not be as
workable as the regular HMA, although it may improve the workability of mixture
produced at the same temperature without any additives. Therefore, the potential old
binder mobilization could be limited if the WMA binders were used.
Figure 4-16 Viscosity testing results
Among the additives, EvoFLEX CA seems more capable of reducing the binder
70
viscosity and the lowest viscosity occurred to binder blended with the combination of 3%
EvoFLEX CA and 0.5 % Evotherm M1. This dosage was recommended by the additive
supplier as the most effective one used for high RAP/RAS mixtures. However, it should
be noted that the EvoFLEX CA addition was 3%, comparatively higher than other
additive addition dosage. The clear reduction in viscosity may be attributed to its high
dosage rate.
Rediset and Sasobit yielded relatively low values of viscosity than Evotherm M1
and Cacebase regardless the product type, but this reduction was very limited. The
different products of the same type of additive did not generate any differences. It was
also noticed that increasing the additive dosage slightly decrease the viscosity level.
Since WMA additive was injected into the GPC system with the binder, the effect
of additive on the GPC results were checked prior to the blending test. The heating effect
was also checked. Figure 4-17 presents the GPC results of binders blended with several
additives as well as the control binder heated at 135oC and 165
oC with tight cap. It can be
seen that GPC testing was not sensitive to the effect of additive or heating for a short
period of time. A one-way ANOVA test was also conducted on the results and the p-value
with α of 0.05 is 0.88. This indicates that no statistical difference can be found within this
group of samples. The limited effect of additive may be attributed to the relatively low
adding dosage.
71
Figure 4-17 Effect of WMA additive or short period heating on GPC results
4.2.3.1 RAP
Figure 4-18 shows the GPC results of binders extracted from coarse virgin
aggregates and medium RAP aggregates. Control samples were prepared for both
foaming and non-foaming mixtures since they were produced with different mixers at
different labs. Only one representative of the same type of additive was reported for
general analysis, since the difference within the same additives was very limited. As
expected, LMS% values of binders extracted from coarse virgin aggregates were lower
than those from medium RAP aggregates, indicating lower RAP binder concentration on
virgin aggregates.
72
(a) GPC results of non-foaming mix (b) GPC results of foaming mix
Figure 4-18 GPC results of coarse-medium RAP mix
According to previous research, a parameter called “blending ratio” was used to
quantify the blending degree, defined as the LMS% ratio of the two binders [22, 39] [Eq.
(4-5)].
(4-5)
The blending ratio proved to be able to quantify the blending level, however,
could be controversial. Imaging there is no blending, the binder coating the virgin
aggregate should be totally from virgin binder. Then the blending ratio can be calculated
as the ratio of LMS% of virgin binder over LMS% of blend of virgin and RAP binders
coating the RAP aggregates. Therefore, the blending ratio is calculated as a non-zero
number under no-blending case. If the similar idea is still used, a more reasonable
calculation method should take out the LMS% of virgin on both sides, which is expressed
as Eq. (4-6).
(4-6)
In this study, the LMS% for virgin binder was obtained as 20.337, thus Eq. (4-6)
in this study can be written as Eq. (4-7).
73
(4-7)
Using Eq. (4-7), the blending ratio for the mixes reported in Figure 4-18 can be
calculated and presented in Figure 4-19. Generally, the blending ratio of non-foaming
WMA is slightly higher than control mix produced at 135oC, but lower than the 165
oC
mix. The foaming mix exhibited higher blending ratio than both control mixes. However,
the difference is not appreciable, especially among WMA mixes.
(a) GPC results of non-foaming mix (b) GPC results of foaming mix
Figure 4-19 Blending ratio of coarse-medium RAP mix
Figure 4-20 and Figure 4-21 present the GPC results of non-foaming and foaming
WMA mix with medium-fine design. Using Eq. (4-7), blending ratio was calculated and
presented in Figure 4-22 and Figure 4-23.
74
Figure 4-20 GPC results of non-foaming medium-fine RAP mix
Figure 4-21 GPC results of foaming medium-fine RAP mix
Figure 4-22 presents the blending ratio of non-foaming medium-fine RAP mix.
75
The effect of the products within the same type of additive was observable, so the figure
was processed with the additives of the same type in the same group. Generally, the
WMA mixes yield blending ratios obviously higher than control mix produced at 135oC,
but still lower than control mix made at 165oC. This finding is consistent with that
obtained from the coarse-medium mixes.
Figure 4-22 Blending ratio of non-foaming medium-fine RAP mix
For Evotherm related additives, adding 3% EvoFLEX CA with or without
Evotherm yielded a blending ratio close to the 165oC control mix. Blending ratio
increased when adding higher dosage of Evotherm. Rediset 1102C mix exhibited a higher
blending ratio than another product 1106 at a dosage of 1.5%. Cecabase with higher
dosage rate also improved blending. Among the four sasobit additives, sasobit yielded the
highest blending ratio. It seems the results from different type of additive agreed on a
trend that increasing the WMA additive increased the blending ratio. Among various
76
products from the same company, no appreciable difference can be found. All these
findings mentioned above indicate that blending remains a critical question over
non-foaming WMA.
Figure 4-23 Blending ratio of foaming medium-fine RAP mix
Unlike the non-foaming additives, the foaming technology was found to increase
the blending ratio of the RAP mix (Figure 4-23). The water content seems have very little
effect. This indicates that foaming WMA may improve the blending between the virgin
and old binder in RAP mix.
4.2.3.2 RAS
Figure 4-24 presents the GPC results of the binders extracted from medium-fine
RAS mix. The blending ratio can be seen in Figure 4-25.
77
(a) GPC results of non-foaming mix (b) GPC results of foaming mix
Figure 4-24 GPC results of medium-fine RAS mix
(a) GPC results of non-foaming mix (b) GPC results of foaming mix
Figure 4-25 Blending ratio of coarse-medium RAP mix
When it comes to RAS, the highest blending ratio was found at 165oC control
mix, with a value merely over 40%. This supports the concern that RAS is very stiff at
normal mixing temperatures and difficult to be mobilized. Only one additive of the same
type was selected to report herein, since the results were fairly close. The mixtures
produced with foaming technology showed lower blending ratios than the control mix,
which is contradicting the finding obtained from RAP mix. This may indicate that
temperature is more important than coating when RAS is used. 135oC seems to be too
low to handle the RAS materials. The blending in WMA-RAS mixtures should be further
evaluated.
78
4.3 Results from Extraction Method
4.3.1 Validation Results of Staged Extraction
Figure 4-26 shows the steel-ball testing results. It can be seen from Figure 4-26
that the percent weight of each layer was controlled within the range 13% to 18%. The 1
minute interval was determined by several trials. According to the GPC testing results
[Figure 4-26(b)], it was found LMS% values of first two layers were approximately the
same with that of virgin binder. The LMS% increased from layer three to inner layers,
with LMS% of last layer close but not exceeding the level of RAP binder. This finding
clearly shows that the composite binder film coating the steel ball was stripped by the
solvent layer by layer. Although this is an ideal and well controlled composite system, the
staged extraction method can be validated.
(a) Percent weight of each layer (b) GPC testing results
Figure 4-26 Steel-ball testing results
79
4.3.2 Effect of Extraction Time
Figure 4-27 presents the GPC results of 1-minute extraction and step-extraction
described above on RAP particles. LMS% values of all the samples are found to be
within the range generated by virgin binder and RAP binder, which follows the common
sense that each layer is a combination of virgin and RAP binders. It can be seen that
LMS% tends to increase from the outmost layer (layer 1) to the innermost layer (layer 6),
regardless of the extraction methods. However, the variation among the layers generated
by 1-minute extraction was relatively smaller, while the step-extraction yields a more
differentiable LMS% distribution. In conjunction with the weight distribution presented
in Figure 5.9, the step-extraction presented in this study seems to generate a more
favorable evaluating system and should be used for further research.
According to the results of step-extraction, it can be found that the first layer is
close to the virgin binder, while the last two layers are similar to the RAP binder. The
LMS% values of layer 2, 3 and 4 stand in the middle, serving as the blending zone with
unknown blending level.
Figure 4-27 GPC results of RAP samples
Figure 4-27 presents the GPC results obtained from extraction of RAS particles.
80
The same finding was observed as that of RAP extraction. This means the step-extraction
proposed in this study can also be extended to RAS blending research.
It is noted that the LMS% of the innermost layer is not very close to the RAS
binder level. This does not mean the virgin binder could blend into the inner layer of RAS
binder, since the fine particles passed No. 8 sieve were RAS particles only and a large
amount of virgin binder might coat the RAS particles with a heavy film. The thick
coating may have brought uncertainties to the system. Further research with more
reasonable design should be conducted to validate this assumption.
4.3.3 Diffusion Studies
Figure 4-28 presents the staged extraction results of coarse aggregates separated
from 50% RAP mixture. According to Figure 4-28(b), the weight of each stripped layer
was similar, indicating the layers may be extracted in the similar film thickness. Figure
4-28(a) shows the LMS% of each layer. There is a slight decrease of LMS% from
outmost layer (1st layer) to 2nd layer, and LMS% changes very little from 2nd to the
innermost layer. This may suggest an approximately homogeneous film coating the virgin
aggregates. A one-way ANOVA test was conducted on the data and a p-value of 0.184
was obtained with α of 0.5, which means no significant difference was found within the 4
layers. It is reasonable that the outmost layer exhibited a slightly higher LMS% since it
was exposed to RAP particles.
.
81
(a) GPC results in terms LMS% (b) Percent weight of each layer
Figure 4-28 Results of staged extraction on coarse virgin aggregates in RAP mix
Figure 4-29 presents the results of fine particles separated from the RAP mix. It
can be seen from Figure 4-29(b) that the extracted layer was well-controlled in similar
film thickness. As expected, the LMS% of the binder layers increased from the outmost
layer to the innermost layer, since the un-mobilized old binder coating the RAP aggregate
considerably contributed to formation of inner layers. However, the interesting finding is
that the first two layers exhibited similar LMS% values to those coating the virgin
aggregates. One-way ANOVA test was conducted on the LMS% of the six layers,
including two outmost layers coating RAP aggregates and all the four layers coating
virgin aggregates. A p-value of 0.56 was obtained, which indicates no significant
difference among the six layers. This finding may support the blending scenario proposed
in Figure 6.3. The virgin binder and mobilized RAP binder may be well blend during
mixing and generate a relatively homogeneous film that subsequently coats the virgin
aggregates or RAP aggregates with some un-mobilized old binder still attaching.
According to Figure 6.10(b), the approximately homogeneous blend film accounted for
around 30% to 35% of the total binder coating the RAP aggregates.
82
(a) GPC results in terms LMS% (b) Percent weight of each layer
Figure 4-29 Results of staged extraction on fine RAP aggregates
Figure 4-30 shows the testing results of binders coating the virgin coarse
aggregates. Unlike the RAP mixture, the binder on virgin aggregates was extracted in an
uneven rate but still within a 10% range. It was also found that LMS% of the outermost
layer was a little higher than the other three that showed similar LMS% values. This may
be because the mobilized RAS binder did not blend well with the virgin binder, thus
attached onto the outmost layer of binder coating virgin aggregates.
(a) GPC results in terms LMS% (b) Percent weight of each layer
Figure 4-30 Results of staged extraction on coarse virgin aggregates in RAS mix
Figure 4-31 presents the results obtained from testing on fine RAS particles. The
average LMS% of the outmost layer that accounts for approximately 20% of the binder
was 24.35, close to the average LMS% of the outmost layer on virgin aggregates, 24.23.
83
Starting from the second layer, the LMS% of the binder was significantly higher. This
finding shows that the binder distribution in RAS mixture may be different from the RAP
mixture. The new blending scenario was proposed and illustrated in Figure 4-32.
(a) GPC results in terms LMS% (b) Percent weight of each layer
Figure 4-31 Results of staged extraction on fine RAS aggregates
Figure 4-32 illustrates different blending scenarios of RAP and RAS mixes. The
blending scenario of RAP was validated, to some extent, by the staged extraction results
of RAP mix. Unlike RAP, the staged extraction results on RAS mix suggested a
composite structure of binder coating the virgin aggregates after mixing. The virgin
aggregate was coated by virgin binder first, then the blend of virgin binder and RAS
binder mobilized during mixing re-coated the virgin and RAS aggregates. Due to the
reduction in temperature or intrinsic difference between RAS and virgin binders, the
binder blend could not enter through the virgin binder layer, thus developing a composite
binder system on virgin aggregates. Since RAS was extremely aged during production
process and service life, only a small portion of RAS binder could be mobilized, therefore
a thick inactive RAS binder layer was left remaining coating the RAS aggregates. The
blending process does not stop when mixing is completed. As shown in Figure 6.7, the
mixture is generally kept at relatively high temperature up to several hours, thus diffusion
starts to occur upon completion of mixing process.
84
.
Figure 4-32 Different blending scenarios of RAP and RAS mixes
The same materials used in blending study were also used in diffusion study for
comparison purpose. According to description of sample 1 to 3 in Table 3-17, the coarse
and fine particles from RAP mix were conditioned in a vacuum oven at 125ºC or 155ºC
for 1 hr to simulate the storage temperature and time for WMA and HMA, respectively.
The lab testing results showed statistically the same LMS% values for binder layers
extracted from the same batch of particles. The authors of this study shortened the testing
time to 15 minutes and the testing results were arranged in Figure 4-33. After conditioned
at 125ºC for 15 minutes, the binder on virgin aggregate tended to be more homogeneous,
with a p-value of 0.636 from one-way ANOVA test, compared to a p-value of 0.184
without diffusion. `
On the RAP aggregates, a fairly homogeneous binder system was generated after
conditioned at 155ºC for 15 minutes, with a one-way ANOVA p-value of 0.604. The
binder exhibited a tendency of being more evenly distributed after treated at 125ºC for 15
minutes, although the p-value of 0.008 indicates that LMS% of all layers were not
statistically the same. This indicates that binder diffusion can be completed during the
mixture storage time for HMA and approximately completed for WMA mixes.
85
(a) Staged extraction on coarse virgin aggregates (b) Staged extraction on fine RAP aggregates
Figure 4-33 Diffusion results of RAP mix
Figure 4-34 presents the diffusion testing results for 10% RAS mix. On the virgin
aggregates, it seems binder diffusion occurred but not in an appreciable level. However,
one-way ANOVA test results show that p-value increased from 0.0149 for sample
without diffusion, up to 0.249 for 125ºC conditioning and 0.119 for 155ºC conditioning.
This may indicate that diffusion occurred, but very slowly. Since the original differences
in LMS% among layers were not significant, the binder on virgin aggregates may be
considered as approximately homogeneous after storage diffusion.
The diffusion concern was noticed on RAS particles. According to the LMS%
results, the homogenization process, especially at higher temperature, could be noticed.
However, the variation of LMS% obtained from the outermost layer to the innermost
layer stayed at a high level. This finding confirmed the concern over binder homogeneity
in RAS mixtures. Since diffusion is highly dependent on temperature, the long-term
diffusion between the virgin and RAS binder at lowered temperature may not be
promising. Therefore, the diffusion homogeneity may remain an issue for RAS mix.
86
(a) Staged extraction on coarse virgin aggregates (b) Staged extraction on fine RAS aggregates
Figure 4-34 Diffusion results of RAS mix
4.4 AFM
AFM imaging was conducted on previously prepared samples made from all the
five binders at 25 ºC. The results are shown in Figure 4-35 to 4-37. According to the
results of topography, the characteristic bee-shaped microstructures can be observed in
both two virgin binders and are found in larger size in binder with lower PG grade. These
bee-shaped microstructures are similar to those found in previous research [31, 34, 35].
On the contrary, no “bees” are found existing in the two tear-off binders, VT and TT.
Both of them show similar topographic images with domains with the size of 1 to 2 µm,
which is largely different from the topography of virgin binder. From the perspective of
profile, these domains look like plenty of round or elliptical “humps” dispersed onto the
surface of a lower homogeneous matrix. It should also be brought into attention that the
topography of binders extracted from tear-offs is significantly rougher than virgin ones.
These observations validate the feasibility of using AFM to differentiate aged RAS binder
and virgin binder through direct detection in solid state. The hump-shaped
microstructures can be seen as the fingerprint of the RAS binder under AFM. These
changes in topography can be attributed to either the polymer added during the shingle
production or significant aging in the air-blown process or after years’ service life.
87
The interesting finding is the typical bee-shaped domains can also be observed in
the topographic images of TM binder, which has gone through the air-blown process for
production. Meanwhile, the roughness of TM binder is also found in between the tear-off
binders and virgin ones. This indicates that post-manufactured RAS binder, in the
microstructural view, behaves like a transition from virgin binder to further aged tear-off
binder. This finding supports the assumption that the microstructural changes, for the
most part, result from aging rather than interaction between polymer and asphalt binder.
Similar to the topography, the phase image of each binder selected in this study
also shows its uniqueness. The phase change transforms from virgin binder to most aged
tear-off binder in the order from simplicity to complexity. Only two apparent phases can
be detected in softest PG 52-58 binder, which is similar to the image of AAN binder
observed by Mason et al. [28]. A third phase is found in PG 64-22 binder, consisting of
flake-like domains [28] separating the dark and light phases. The TM binder shows less
comparable domains than virgin one, while the dark phase is almost invisible in
morphological images in both two tear-off binders. The phase images, therefore, proved
to provide an alternative to fingerprint the RAS and virgin binders.
88
Figure 4-35 Topography and phase images of PG 52-28 and PG 64-22 virgin binder
89
Figure 4-36 Topography and phase images of TM binder
90
Figure 4-37 Topography and phase images of TT and VT binder
91
4.4.1 Temperature Dependence of Microstructures in RAS Binder
One type of tear-off binder (TT) was made into samples and scanned through
AFM after treatment in different temperature. As can be seen in Figure 4-38, the sample
was heated to different target temperature in a 20 ºC step, hold for 5 minutes, and then
cooled to 25 ºC for AFM imaging. The maximum temperature was set at 180 ºC that is
the highest temperature most asphalt plants can reach. The observations at 120 ºC and
180 ºC are reported only in this study for better analysis.
Figure 4-38 Schematic of thermal conditioning of TT binder
Figure 4-39 demonstrates the topographic images of TT binder after treatment at
selected temperatures. The bright hump-shaped microstructures diminish in size with the
treating temperature increasing up to 80oC. The most dramatic change occurs from 40 ºC
to 60 ºC. The topography of this specific binder changes very little after the temperature
exceeds 80 ºC, where the hump-shaped microstructures are found to “melt” with small
light nuclei still remaining dispersed in the melted phase. The images vary little from 80
ºC to 180 ºC.
92
Figure 4-39 Topographic images of TT binder after temperature treatment
4.4.2 Blending Degree between RAS and Virgin Binder
A two-layered sample sketched in Figure 4-40 was designed to evaluate the
blending degree between RAS and virgin binder. AFM images acquired from locations on
the top of the sample were used to investigate whether virgin binder from the bottom
layer could blend through to the surface of the upper layer of RAS binder, after being
treated under certain time and temperature correlated to plant production. The RAS layer
was designed to be smaller in size to intentionally make an interfacial zone that was also
scanned in this study, in order for better understanding on blending.
93
Figure 4-40 Experimental design of observing the blending degree between RAS and
virgin binder
Figure 4-41 presents the representative topographic AFM images acquired from
scanning on the top of the sample. The sample was treated at 180 ºC for 5 minutes, which
is theoretically long enough to ensure complete blending occurring. It can be seen that the
topography on the surface of the layered sample is nothing different from the TT RAS
94
binder, regardless of the binder grade of the bottom virgin layer. This indicates that the
virgin binder selected in this study could not blend through to the surface of the tear-off
RAS binder layer around 300 µm in thickness. Since blending between virgin and aged
binder was found to be a function of treating temperature and time, the observation of
no-blending may be attributed to limited time. This assumption motivated increasing the
treating time to a fairly long 30 minutes. The topography of the layered samples,
however, changed very little. Accordingly, it seems of significance to detect what happens
on the interfacial zone for a deeper understanding of blending between RAS and virgin
binder.
95
Figure 4-41 Comparison of scanning results on the top of the layered sample
: (a) Top layer TT binder and bottom layer virgin bitumen PG6422; (b) Top layer TT
binder and bottom layer virgin bitumen PG5228; (c) Control TT binder
96
4.4.3 Scanning on the Interfacial Zone
As can be seen in Figure 4-42, the TT-RAS layer was made smaller in diameter
and spread onto the bottom virgin binder at ambient temperature. The probing was
conducted alongside the interfacial zone that can be easily differentiated by the optical
microscope installed in AFM.
Figure 4-42 Probing the interfacial zone
In this paper, the AFM images acquired from scanning on the interfacial zone
between TT RAS and PG 52-28 binder were reported to address the detailed observation
on blending. As can be seen in Figure 4-43, both “bees” and “humps” can be found in
topographic images of the interfacial zone, which are typical microstructures representing
virgin binder and RAS binder, respectively. It seems the two binders were mixed but not
blended into one ‘new’ material, indicating a poor compatibility between the virgin
binder and significantly aged tear-off RAS binder. The mixing may be attributed to either
upward or lateral movement of the virgin binder.
The mixing zone is found to be around 25 to 30 μm. However, the size of the
mixing zone observed in this study does not indicate the real mixing scale occurring in
the drum, since the film thickness on the edge of the top RAS layer was not precisely
97
determined.
98
Figure 4-43 Comparison of scanning results on the interfacial zone of the layered sample
and detailed description of mixing zone
The co-authors of this paper recently published a study to address the RAP-virgin
blending under AFM for the first time [31]. Two blending scenarios were proposed in
terms of microstructural properties (Figure 4-44). Scenario A refers to a merely mixing of
two distinct colloidal fluids, according to which both the colloidal particles can be found
in the mix. On the contrary, a complete blending that generates a new “colloidal” material
is also possible and illustrated as scenario B.
Figure 4-44 Blending scenarios (adapted from [31])
On the basis of the observation of RAP-virgin blending, as illustrated in Figure
4-45, the authors [31] concluded that “obviously scenario B is closest to the observations,
hence one should speak about blending rather than mixing”.
99
Figure 4-45 Phase image of RAP-virgin binder blending (adapted from [31])
The observations of RAS-virgin binder blending obtained in this study (Figure
4-43), interestingly, are more likely to correspond to scenario A, which is mixing rather
than blending. This observation may lead to concerns over the binder segregation of RAS
mixtures. Meanwhile, not all the characterizing tools used for RAP binder/mixture can be
easily extended to RAS, due to the difference in nature between RAS-virgin blending and
RAP-virgin blending.
100
CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS
A series of studies were conducted to address the blending issues in warm and hot
mix asphalt containing recycled asphalt pavement (RAP) and recycled asphalt shingle
(RAS). The blending was evaluated in terms of analysis on old binder mobilization,
binder homogeneity and diffusion. Several new methods were developed to characterize
the blending process. Based on the results obtained from the aforementioned studies, the
following recommendations can be made:
1) A strong correlation existed between the percentage of LMS and G* of asphalt
binder based on the comparison of GPC and DSR test results. As mixing time
and temperature increased, more blending occurred in the RAP/RAS mixture.
The size of virgin aggregate did not affect the blending efficiency of RAS in
pavement mixtures. The most efficient blending of RAS may occur at
approximately 5% RAS content.
2) A new parameter, large molecular size percentage (LMS%) related to
molecular weight distribution derived from gel permeation chromatography
(GPC) analysis, was developed to differentiate the RAP/RAS and virgin
binders as well as their blends. “Blending Charts” could be generated between
the RAP/RAS binder content in the blend with the newly defined LMS% and
the relations were found to be linear. The RAP/RAS binder mobilization rate
defined in this study could be determined by LMS% analysis of binders
extracted and recovered from the virgin aggregates after mixing in the
laboratory, with the use of the “Blending Charts”. The results show that RAP
binder mobilization rate decreased with the increase of the RAP percentage in
the mixture with mobilization rates close to 100% at low RAP mixtures (10%
and 20%), but dropping from 73% to 24% with RAP percentage varying from
101
30% up to 80%. RAS binder mobilization rate increased with RAS percentage
growing from 2.5% to 5%, but decreased with RAS percentage passing 5%.
The highest mobilization rate was around 61% and found on 5% RAS mixture
while the mobilization rate of mixture containing 10% RAS could be as low as
36%.
3) The feasibility of using staged extraction was validated in this study. It was
found TCE was the most effective solvent used in the study for staged
extraction that dissolved the asphalt binder without preferential dissolution.
Meanwhile, TCE was found to have the highest dissolving rate. The binder
coating on the raw RAP and RAS aggregates was proved to be homogeneous
and the layer stripping did occur in a well-controlled composite binder system.
A well designed step-extraction method with progressive wash times could
replace equal-time extraction method, and yielded better analysis. Partial
blending was observed within the coating of RAP particles, while the
RAS-virgin blending on RAS aggregates should be further evaluated.
4) Based on well-designed staged extraction and GPC analysis, it was found that,
in RAP mix, binder film coating virgin aggregates was approximately
homogeneous, while a non-homogeneous binder was generated on RAP
aggregates. The model of binder blend coating the virgin and RAP aggregates
with inactive RAP binder still attaching was validated in this study. A
potential composite binder system was found coating the virgin aggregates in
RAS mix. The diffusion study shows that within the mixture storage time,
binder diffusion can be accomplished in both warm and hot mixes containing
RAP, indicating old binder mobilization, rather than binder homogeneity,
could be more critical in RAP mix. The binder diffusion in RAS mix was
captured in a very slow rate. It was suggested that old binder activation and
binder homogeneity can both be critical for RAS mix.
102
5) WMA additives slightly decreased the viscosity of the asphalt binder at 135oC.
However, Binder tested at 165oC showed significantly lower viscosity than
WMA binders. This may raise the concern over workability of the WMA mix.
WMA additives yielded higher blending ratio than control mix produced at
135oC, but the temperature of 165
oC still produced the mix with the highest
blending ratio value. This indicates that a concern still exists over asphalt
blending even if WMA additives are used. Foaming technology yielded a
higher blending ratio, indicating foamed WMA may yield a higher blending
than regular HMA. It was also found that temperature rather than coating is
more critical in RAS blending. Finally, the mix produced with coarse virgin
aggregates and medium RAP may not be sensitive enough to test the effect of
WMA additives on blending, while the mix with medium virgin aggregates
and fine RAP was more effective.
6) AFM could be used to characterize microstructural properties of the selected
virgin, post-manufactured RAS and post-consumer RAS (tear-off) binders, as
well as the temperature dependence of microstructures in one type of tear-off
RAS binder. The blending of virgin-RAS binder was first observed in this
study. According to the observations, AFM proved to be capable of
differentiating virgin binder from RAS binder in terms of microstructures. The
microstructures of tear-off RAS binder was found to be
temperature-dependent, but changed very little within the range from 60oC to
180oC. Virgin binders selected in this study could not blend through a RAS
binder layer of 300 μm within 30 minutes at 180oC. On the basis of
observations on the interfacial zone, RAS binder was found to be “mixing”
but not “blending” in a mixing zone of 25 to 30 μm.
103
Recommendations
On the basis of the conclusions obtained in this study, the following
recommendations can be made:
1) Better tracking materials, other than the round aggregate could be used for
the quantitative evaluation of the old binder mobilization rate. The texture,
size and other surface properties of the tracking materials could be
considered as influence factors.
2) It is recommended to develop new methods to quantify the binder
homogeneity through AFM. Statistical methods could be used to track the
numerical change of the domains. Layered system with well controlled
binder film thickness is recommended to characterize the diffusion
coefficient of the virgin binder through the old binder.
3) Neutron scattering are also recommended for use in blending research.
Samples with different blending degrees may express different inner
structural properties and might be revealed by neutron scattering detection.
Additionally, neutron scattering samples will not go through any destructive
damage during preparation and testing processes.
4) Test methods designed in this research have been proved to be useful in
laboratory research. However, their effectiveness in field research still need
104
to be further verified. In the follow-up study, the test methods for blending
efficiency need to be modified based on the requirement of field research.
105
APPENDICES
106
APPENDIX A: GPC Chromatograms of Test in Figure 3-2
107
108
109
110
111
112
113
114
115
116
117
118
119
APPENDIX B: GPC Chromatograms of Test in Figure 3-4
120
121
122
123
APPENDIX C: GPC Chromatograms of Test in Figure 3-5
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
APPENDIX D: GPC Chromatograms of Test in Table 3-5
142
143
144
145
146
147
148
149
150
151
152
APPENDIX E: GPC Chromatograms of Test in Table 3-8
153
154
155
156
157
158
159
160
APPENDIX F: GPC Chromatograms of Test in Figure 4-1
161
162
163
164
165
166
167
168
169
170
171
172
REFERENCES
1. Huang, B., et al., Laboratory investigation of mixing hot-mix asphalt with
reclaimed asphalt pavement. Transportation research record: journal of the
transportation research board, 2005. 1929(1): p. 37-45.
2. McDaniel, R.S., et al., Recommended use of reclaimed asphalt pavement in the
Superpave mix design method. NCHRP Web document, 2000. 30.
3. Kandhal, P.S., K.Y. Foo, and R.B. Mallick, A critical review of VMA requirements
in Superpave. 1998, National Center for Asphalt Technology.
4. Soleymani, H.R., et al., Investigation of the black rock issue for recycled asphalt
mixtures. ASPHALT PAVING TECHNOLOGY, 2000. 69: p. 366-390.
5. McDaniel, R.S., et al., Effects of reclaimed asphalt pavement content and virgin
binder grade on properties of plant produced mixtures. Road Materials and
Pavement Design, 2012. 13(sup1): p. 161-182.
6. Bonaquist, R., Can I run more RAP? HMAT: Hot Mix Asphalt Technology, 2007.
12(5).
7. Mogawer, W., et al., Performance characteristics of plant produced high RAP
mixtures. Road Materials and Pavement Design, 2012. 13(sup1): p. 183-208.
8. McCann, M., J.F. Rovani, and K.P. Thomas, Instrumental method suitable for the
detection of polymers in asphalt binders. TRB Rep. 08, 2008. 907: p. 1-14.
9. Lu, X. and U. Isacsson, Effect of ageing on bitumen chemistry and rheology.
Construction and Building Materials, 2002. 16(1): p. 15-22.
10. Siddiqui, M.N. and M.F. Ali, Investigation of chemical transformations by NMR
and GPC during the laboratory aging of Arabian asphalt. Fuel, 1999. 78(12): p.
1407-1416.
11. Kim, K.W. and J.L. Burati Jr, Use of GPC chromatograms to characterize aged
asphalt cements. Journal of materials in civil engineering, 1993. 5(1): p. 41-52.
12. Churchill, E.V., S.N. Amirkhanian, and J.L. Burati Jr, HP-GPC characterization
of asphalt aging and selected properties. Journal of materials in civil engineering,
1995. 7(1): p. 41-49.
13. Lee, S.-J., et al., Aging analysis of rubberized asphalt binders and mixes using gel
permeation chromatography. Construction and Building Materials, 2011. 25(3): p.
1485-1490.
14. Snyder, L.R., Determination of asphalt molecular weight distributions by gel
173
permeation chromatography. Analytical Chemistry, 1969. 41(10): p. 1223-1227.
15. Gilmore, D.W., et al., Determination of the presence of anti-stripping agents in
asphalt cement by high-performance liquid chromatography. Properties of
flexible pavement materials, ASTM STP, 1983. 807: p. 46-54.
16. Kim, K.W., et al., Estimation of RAP’s binder viscosity using GPC without binder
recovery. Journal of materials in civil engineering, 2006. 18(4): p. 561-567.
17. Shen, J., S.N. Amirkhanian, and S.-J. Lee, HP-GPC characterization of
rejuvenated aged CRM binders. Journal of Materials in Civil Engineering, 2007.
19(6): p. 515-522.
18. Kim, H., et al., Quantification of oxidative aging of polymer-modified asphalt
mixes made with warm mix technologies. Journal of Materials in Civil
Engineering, 2012. 25(1): p. 1-8.
19. Jennings, P., et al., High pressure liquid chromatography as a method of
measuring asphalt composition. 1980.
20. Jennings, P. and J. Pribanic, The expanded Montana asphalt quality study using
high pressure liquid chromatography. 1985.
21. Yapp, M.T., A.Z. Durrani, and F.N. Finn, HP-GPC and asphalt characterization
literature review. 1991.
22. Zhao, S., et al., Characterizing rheological properties of binder and blending
efficiency of asphalt paving mixtures containing RAS through GPC. Journal of
Materials in Civil Engineering, 2013. 26(5): p. 941-946.
23. Daly, W.H., I. Negulescu, and S.S. Balamurugan, Implementation of GPC
Characterization of Asphalt Binders at Louisiana Materials Laboratory. 2013.
24. Loeber, L., et al., New direct observations of asphalts and asphalt binders by
scanning electron microscopy and atomic force microscopy. Journal of
Microscopy, 1996. 182(1): p. 32-39.
25. Loeber, L., et al., Bitumen in colloid science: a chemical, structural and
rheological approach. Fuel, 1998. 77(13): p. 1443-1450.
26. Pauli, A., et al., Atomic force microscopy investigation of SHRP asphalts: Heavy
oil and resid compatibility and stability. Preprints-American Chemical Society.
Division of Petroleum Chemistry, 2001. 46(2): p. 104-110.
27. Jäger, A., et al., Identification of four material phases in bitumen by atomic force
microscopy. Road Materials and Pavement Design, 2004. 5(sup1): p. 9-24.
28. MASSON, J.F., V. Leblond, and J. Margeson, Bitumen morphologies by phase‐detection atomic force microscopy. Journal of Microscopy, 2006. 221(1): p. 17-29.
29. Pauli, A., et al., Morphology of asphalts, asphalt fractions and model wax-doped
asphalts studied by atomic force microscopy. International Journal of Pavement
Engineering, 2011. 12(4): p. 291-309.
30. Das, P.K., et al., Micromechanical investigation of phase separation in bitumen by
combining atomic force microscopy with differential scanning calorimetry results.
174
Road Materials and Pavement Design, 2013. 14(sup1): p. 25-37.
31. Nahar, S., et al., First observation of the blending zone morphology at the
interface of reclaimed asphalt binder and virgin bitumen. Transportation Research
Record, 2370 (1), 2013; Authors version, 2013.
32. Wu, S.-p., et al., Influence of aging on the evolution of structure, morphology and
rheology of base and SBS modified bitumen. Construction and Building Materials,
2009. 23(2): p. 1005-1010.
33. Tarefder, R.A. and A.M. Zaman, Nanoscale evaluation of moisture damage in
polymer modified asphalts. Journal of Materials in Civil Engineering, 2009. 22(7):
p. 714-725.
34. Nahar, S., et al., Temperature and thermal history dependence of the
microstructure in bituminous materials. European Polymer Journal, 2013. 49(8): p.
1964-1974.
35. Nahar, S., et al. Is atomic force microscopy suited as tool for fast screening of
bituminous materials? An inter-laboratory comparison study. in 92nd Annual
Meeting Transportation Research Board, Washington, USA, 13-17 January 2013;
Authors version. 2013. Transportation Research Board.
36. Garcıa, R. and R. Perez, Dynamic atomic force microscopy methods. Surface
science reports, 2002. 47(6): p. 197-301.
37. Eaton, P. and P. West, Atomic force microscopy. 2010: Oxford Univ. Press.
38. Shirodkar, P., et al., A study to determine the degree of partial blending of
reclaimed asphalt pavement (RAP) binder for high RAP hot mix asphalt.
Construction and Building Materials, 2011. 25(1): p. 150-155.
39. Bowers, B.F., et al., Investigation of reclaimed asphalt pavement blending
efficiency through GPC and FTIR. Construction and Building Materials, 2014. 50:
p. 517-523.
40. Eddhahak-Ouni, A., et al., Experimental investigation of the homogeneity of the
blended binder of a high rate recycled asphalt. Road Materials and Pavement
Design, 2012. 13(3): p. 566-575.
41. Stroup-Gardiner, M. and J. Nelson, Use of normal propyl bromide solvents for
extraction and recovery of asphalt cements. National Center for Asphalt
Technologies, Report# NCAT, 2000. 6: p. 2000.
42. Bowers, B.F., Investigation of Asphalt Pavement Mixture Blending Utilizing
Analytical Chemistry Techniques. 2013.
43. Zearley, L.J., Pénétration characteristics of asphalt in a recycled mixture. 1979.
44. Noureldin, A.S. and L.E. Wood, Rejuvenator diffusion in binder film for hot-mix
recycled asphalt pavement. Transportation research record, 1987(1115).
45. Zhou, F., J.W. Button, and J.A. Epps, Best Practice of Using RAS in HMA. Vol. 7.
2012: Texas Transportation Institute College Station.
46. Karlsson, R., U. Isacsson, and J. Ekblad, Rheological characterisation of bitumen
175
diffusion. Journal of materials science, 2007. 42(1): p. 101-108.
47. Rad, F.Y., Estimating blending level of fresh and RAP binders in recycled hot mix
asphalt. 2013, University of Wisconsin Madison.
48. Kriz, P., et al., Blending and diffusion of reclaimed asphalt pavement and virgin
asphalt binders. Road Materials and Pavement Design, 2014. 15(sup1): p. 78-112.
49. McGraw, J., et al., Recycled asphalt shingles in hot mix asphalt. Asphalt Paving
Technology-Proceedings, 2007. 76: p. 235.
50. Xiao, F., V. Punith, and S.N. Amirkhanian, Effects of non-foaming WMA additives
on asphalt binders at high performance temperatures. Fuel, 2012. 94: p. 144-155.